Python OpenCV: Mastering Computer Vision with 10 Hands-On Projects

1. Class Intro: Hello. Welcome to our class Python Open City Mastering Computer Vision. My name is John the Shortcut. I am FOSTEC web developer, Python programmer at Online Instructor, and I'm going to instruct you through the whole class. I am thrilled to be your guide on this journey into the world of computer vision and image processing. In this course, we will deep dive into the Open C, the industry standard library that bring the magic of computer vision to Python. By the end of this course, you will not only understand the fundamental of OpenCV, but you will also gain practical skill that you can immediately apply thanks to our ten hands on project. Now, let's see what exactly we are going to learn from here. We will start with the basics. First, we are going to learn how we can install OpenCV in our local machine. Then we are going to read image, resizing and rescaling frames. We are going to draw different shape and put text on image. Then we are going to learn important functions. Next, we are going to learn transition, rotation, translation, conduit detection. Color spaces, color channels, blurry, Btis operation, masking, gray scale histogram, color image histogram and thresholding. After learn all of the basic concept, we are going to jump into the project section. We are going to conduct multiple real world projects, such as text detection and extraction using OpenCV and draw rectangle shape and extract object, black dot white dot detection. Tis pin coordinates when we click on image. Then we are going to jump into the most important part, which is invisible T shirt project, Background substriction, phase detection using OpenCV. Also, we're going to contact another project detecting objects of similar color in Python. And this is not the end. Time to time, we are going to update this class, and we are going to add new project in this class, such as finger detection, eye detection, control brightness using fingers and many more. Every week, I'm going to add new concept on it. By the end of this class, you will have the skill to build your own computer vision projects and automate your task. After this course, you can explore area like AI and argument reality. So what are you waiting for? Let's start together. 2. Installing openCV in our local machine: Hey, this is the first tutorial related Python OpenCV course. Hell. My name is John Shoker. I host a web developer and online instruct. We are going to start out tutorial by installing OpenCV in our local computer. So let's jump into the computer screen. So as you can see, we are in my computer screen, and I open Windows Power Shel. Basically, OpenCV is a computer vision library. It is available in Python, C plus plus, and Java language. A computer vision is a application of deep learn. And it primarily focusing on media files. It could be images, otherwise, video files. I am assumed that you are already installed Python in your computer. And to check, is it installed in your local computer or not, just type Python space Virgin. Then hit Enter. After hit Enter AsimCC it return Python version 3.11 0.4. Just make sure you have the latest version of Python, otherwise, greater than 3.7. So I hope you are already installed Python in your computer. But now I'm going to install the packages that is required for this course. So I'm going to install our main package OpenCV to install OpenCV in your system, just to type PP command. P, install OpenCV Ctrip. Dash Python. Sometimes people may tell you just download OpenCV dash Python. Basically, Opens dash Python is the main package. It is the main module, but Opensi Ctrip contained main module with that it contain contribute modules, which is provided by the community. So I'm going to hit Enter and waiting for download. As you can see, requirement already satisfied. I already download this module, but you need to complete your download process. With that, also, I'm going to update my P version. So I'm going to copy this command and paste it here. Then I'm going to hit Enter. It's going to update my Pin. So I successfully update my P version. But if you notice, you can see, when I try to download OpenCV, it also download Numpie warning. Numbi is kind of scientific calculation packaging Python. It is extensively using matrix, manipulation, et cetera. Maybe we need to use Numpy in some of this tutorial, but don't worry about it because it is simple and relatively get easy to start. So this is it for this tutorial. In this tutorial, I just show you how we can download Open C. In the next tutorial, we are going to learn how we can read images and video using OpenC. So thanks for watching this video, stay tuned for this tutorial. 3. Reading images and video using openCV: Hey, everybody. Welcome back to another field. In this video, we are going to learn how we can read images and videos using Python Open Cit. So let's jump into the isa Studio code editor. As you can see, here I open my visa Studio code editor. And at the website, you can see this is my current working ding. At first, in my current working directory, I'm going to create a Python five, and I'm going to name it main dot py main dot py. At the first half of video, we are going to talk about how we can read images. And next we're going to talk about how we can read videos. So at first in our Python file, we need to import OpenCV, so type, import CV two as C. And to read an image in OpenCV, we need to use imaged meth we need to type cd dot Image R. IMR. Then inside the rounds inside the quotation, we need to provide the file path. As you can see in my working directory, there is another folder named images. And if I open it, here you can see, we have to tap two image, cat one dot JPG cat dot JPG. So at first, I'm going to read cat dot JPG. So inside the quotation, first, I'm going to tag the folder name Images. Images slash I'm going to target the picture, cat dot JPG. Also, I want to capture this image in a variable, and our aria is ING. ING equal to Cv dot image root function. Now, once we read this image, we need to show this image. To show this image, we need to use a method, which is image show, some type cv dot image show. IN show. Then inset the round presses, need to pass total two parameter. This method going to display the image in a new window. So first, here we need to provide the name of the new window, and the Hemotype can. Next, we need to pass the matrix, the image we read. So I'm going to type IMG, this variable, and I'm going to set this spot. Now I want to run a keyboard binding function before I run this scope because here we need a specific d so I'm going to call another method, which is known as wet key. Swing dot wet key. This one. And instead the round presses, I'm going to pass Z. It is a keyboard binding function. It is wait for specific. Until I press any key of mil keyboard. If I pass zero, it's going to wait for infinite time. Basically, I want to say until I press any key from our keyboard, it's not going to remove the image. Let's run the code in our terminal. So I'm going to set the file and after set the file, I'm going to hit the run button. After press the Run button, as you can see, it return the image. Here it open window and display the image in this window. And if you want to kill this window, you can praise any key from your keyboard. Otherwise, you can click on this close icon. So I'm going to praise space key. So as you can see, it kill the window. This is cool. Now, the image I open from my current working directory, it's a small resolution image. Let me show you. As you can see in my current working directory, there is two image. Get one dot JPG and Go dot JPG. Both the images are same, but their resolution is different. If I show you G dot JPG, now you can see its dimension is 604 13. But if I Ober my cursor, get one dot JPG, now you can see its dimension is much more bigger than previous one. It is 5,026 y 3,458. Now I'm going to open the same cat image, but this time, I'm going to open the larger version. So let's get to the salt studio code. I'm going to change the image name. I'm going to make it cat one dot GPG, and I'm going to step this file, and I'm going to run this code. And I'm going to hit the Run button. After hit the run button, as you can see, the image goes way off screen. So I'm going to close this image by pressing any key of my keyboard. I'm going to press B. After press B, it kill the window. This image is much larger than my screen resolution. Currently, obviously we don't have any in build resizing techniques that work according to the computer resolution, but we can resize the image. In the next tutorial, I'm going to talk about it. So if you have large image, possibly, it can go off screen. In this tutorial, just focus on reading images and videos. So this is for reading images. Now, let's focus on how we can read videos. But before I'm going to select all the code and I copy it. Then I'm going to create another fine in my working directory, which is main two dot Pine. And I'm going to paste all the code from my previous fine. This time, we are going to read video five. So at first, I'm going to comment out these lines. I don't need it for now. Now, here's why I show you my current working directory, there is another folder named Videos. If I open it, there is a video, video one.nb4. And to read a video, here I'm going to create a variable capture. Equal to here I'm going to use video capture method. Cv dot, video catureT one. And rounds. This method either take Itiser value like one, two, three. Otherwise, we can pass a file path. Now you might a portion why we need to pass in dieser value. Because using integer value, you can connect with your webcam. If you have one webcam, you can pass zero. Most of the cases you webkem reference by integer zero. If you have multiple one, then you can change the number. But before, at first, I'm going to read the existing video from our file path. I said the single quotation. First, we need to provide the folder name, some type videos then inside this videos folder, we have this video one. So Intro select video one.b4. Remember, reading videos is different from reading image. We need to read videos frame by frame. That's why we need to use Hilo. So here, I'm going to use Hilo. Wild, true. Then inside the While loop is crew Comma, frame equal to capture dot read function. Capture dot read function. This capture dot read function going to read this video frame by frame. It read the frame and this wollen value going to tell that it is successful you work or not. To display the video, we actually display the individual frames. So after read every frame of this video, we need to display the frames. To display these frames, we are going to use image show function. So type Cv dot image show is at the round presses, as you know, first, we need to pass a window name, which is video. Then I'm going to pass the actual frames. So I'm to pass this variable frame. So copy the variable name, and I'm going to paste it here. For now, I don't need this wet key function, so I'm going to remove it. Yes, we need wet key, but we're going to use it with p condition. So her type, I v dot wet key, inside the roundresss held pass twin and zero X F equal, equal to b, inside the round resssOd of B, then it's going to kill you into. Then inside the IP condition, I want to break this VI loop. Once we've done outside the I loop, we need to release the capture pointer. So we need to use release functions on type captured Dot, release function, release. Also, we need to destroy all the windows. He tie, CV, dot, destroy all the windows. This one. This is it. Now recap the port. The capture variable is instance of video capture gloss. Then instead the one loop, we read the video frame by frame. Then we displayed every frame using IMS method. If CVD 820 and zero xFF equal to org, D, then break the load. So if I press B, then it's going to stop show video frames. Basically I want to say, if I press D, then it's going to break the load. Then we release the capture device and destroy all the windows. Now, let's set the file and run the code. After I run this code, as you can see, it play our video. And if I praise D, it close our window. Now, if you want to open your webcam, yes, you can. Just do begin this line and comment on previous one and change the video capture video. I want to use zero because I have just only one camera connected with my computer. After step this file, if I run this code and wait for some segins Now you can directly capture your video using OpenCV and you can read it off. So if you want to close this window, just you need to press B command. If you press any other key, it's not going to close your window. In our condition, here we say, D, which going to break the loop? Now, let me show you an important thing. So first, I'm going to kill my terminal, then I'm going to comment out this line and uncomment this one. And I'm going to sub this file. After set this file, I'm going to run this video again. This time, you can notice after complete the video, it's going to show an error in our terminal. So after 10 seconds, after complete the video, as you can see in my terminal, there is an hero. It's mean when our video is over, otherwise stopped, then it's going to show the hero. If there is no frame in this video, it can return hero. Specifically, it return this error minus 215 assertion feel. It's mean obviously could not find any media file at that location. It's going to return the same type of error with image also. Let me show you. So I'm going to kill my terminal and back to main dot Pi file. If I change the file path, if I make it cat three, as you know, there is no file available in our directory cd three. If I set this file and run this command, as you can see in my terminal, it sit on the same error, negative 215 assertion fade. It's exactly show the same error as like video. It could not find any frame. So that is very much for this video. In this tutorial, we learn how we can read image. Also we learn how we can read videos. With that, also we learn how we can read videos through our webcam. So this is it for this tutorial. At the next tutorial, we are going to learn how we can resize and rescale any frames. It could be image, otherwise video. So thanks for watching this video Stune for the next tutorial. 4. Resizing and Rescaling Frames: Hey, guys, good to see you back. Once again, I'm back with another tutorial related open ct. And in this tutorial, we are going to learn resizing and rescaling. So as you can see, we are jumping to the user Studio code editor. In this section, we are going to learn how we can resize and rescale video and images using open C. To rescale a video frame or image, we need to create a function, and our function name is Dave scalFrame. Then inside the round less, here we need to pass some parameter, frame to V size and the scale will scale. Equal to here, we need to pass scale value, how much we want to scale. For now, I'm going to set 0.75. Then inside the rescal function, I'm going to set with W equal to frame dot shape, instead of the squared ss, I'm going to pass one. Then I'm going to multiply with scale. And next, I'm going to duplicate this line and replace with with height. And inside the square versus I'm going to pass zero. Frame dot shape one, basically with of your image and frame dot shape zero specify the height. And as you know, width and height are indiser, so we need to convert it into a indiser. I'm going to use int function. It, and we need to use round *** also going to do the same thing for height in round brass. And now I'm going to declare a variable called dimension. Dimension equal to, and I want to create a tuple. So instead the rounds, I'm going to pass what coma, and our next value is height. And next, I'm going to return v dot resize function. So I'm going to return return, Cv dot, resize. Inside the run press, first, I'm going to pass frame. Next, I'm going to pass dimension. Comma intercolon value equal to Cv dot Enter area. Enter area. Civ dot resize is a open CI function. It is used to resize an image, and frame is our image, what we want to resize. And dimension specify the new size for the image, and it should be a tepel which going to represent height and width. And this parameter specify the interpolition method used to resize the image. Interpolution method determine how pixel values are calculated when an image is resized. CVI Enter area is a specific interpoltion method, optimized for shrinking images. It may result in better quality than other methods when reducing the image size. Don't worry. We are going to learn more about it later. Now I want to read a video five, so I don't need this one. So to revoke this one and back to mean dot Pi five. From this section, I'm to copy this portion. And back to rescale dot Pi five, and I'm going to paste it here. For now, I don't need this line because I don't want to use my webcam. Now, after I read this frame in our VLU, I want to resize the frame. So we're going to call this function. But before, I'm going to declare a variable, and our variable name is frame resize. Frame under sco resize. Equal to, I'm going to call this function, rescale frame. So I copy the function name and I'm going to paste it here, then inside the round dresses and inside the round presses, we need to pass this variable frame. Then we need to show this variable using image show method. So I'm going to duplicate this line. And first, I'm going to name this window, which is video resized. And then we need to pass the frame resized variable. So I copy this variable and repress frame with it. And I'm going to set this file. And before I run this code, I'm going to remove this two line. I don't need this anymore. Otherwise, I can comment out this one. And I'm going to set this file again. After I set this file, if I run this command, it's going to open total two into. First, it's going to open original window, then it's going to open the resized window. Let me prove you that. So after set this file, if I run this code in my terminal, as you can see, it's going to open total two Window. This is the original window with high resolution, and this is the scale window with low resolution. So this is how it's work. This video is resize by 0.75%. If you want to reduce it more, just you need to reduce the value. By default, this function every time going to scale down 0.75%. But if you want to pass your own value, just you need to pass comma had and provide scale value scale 0.20%. And if I run this code, here you can see this is our original size, and this is the decisive version. Now, it's scaled down 20%. So I'm going to praise D to D the window before it ran out of frame. Using the same function, you can resize image. Let me show you how we can resize an image. But before, I'm going to comment out these lines. Somebody use multi lein comment to comment out all the lines. Then I create a variable, and our variable name is resized image. Underscovimage. Resis imaged equal to here I'm going to use the function recess frame. So I'm going to copy the function name and I'm to call the function. Instead the round process, we need to provide the image which I want to show. So I'm going to comment out these lines again, and I'm going to pass this variable in this function. Image. And then I'm going to use I'm Show function. I copy this line and paste it here. First, I'm going to change the Window name, which is resize asco C. And then I'm going to call cv dot wetKeyFunction. Cv dot weit key. Inside the Rundress, I'm going to pass zero. That's it. Before I set this file, and run the code, I didn't mistake. Here, we need to pass this variable. Resized image. And again, I'm going to set this file and run this code. Up to run this code, as you can see, this is the original size of this image, and this is the scale version of this image. We scale down the image window successful. So using same function, we can scale images. Also we can scale videos. Remember, this method is work for images, video and light videos. So using one function you can do all tics. So this is it for this tutorial. At the next tutorial, we're going to learn how we can draw shape using OpenCL. 5. Draw different shapes and put text on image: Hey, everyone, and welcome back to another ID. In this tutorial, I'm going to talk about how can we draw and write images? So let's jump into the is a studio code editor. So at first, I'm going to create a Python file. So here I'm going to type, draw dot py. And then I'm going to import our libraries. At first, I'm going to import CV Import, CV two as CV. Then I'm going to import NumPi as NP. Import, Num Pi, as N P. Also you can see, I already put that image in my current wing dy cat JPG. Then I'm going to read this image using INRdFunction. So here type first hemo crative variable IMG equal to, I'm going to use C dot image read. IaRad. Then the round brass, I at the quotation. We need to provide the file path, CAT two dot JPG. Then we need to display this image using I show function. So I will type cv dot image show IM, so Is the roundreF we need to assign a name to this window, which is C. And next, I'm going to pass the variable, which is ING image. And then we need to use Wet kwatho v dot wet key, wet key, and inside the round presses, I'm going to pass Z. I'm going to set this file. There are two ways to draw on images. You can draw anything in a standalone image like this one. Also, you can create a blank image. So I decide on where to create a blank image. Here types blank, equal to, I'm going to use tt zeros. Then the round brass, I'm going to pass resolution 500 by 500. Also, we need to put it in a round brass. They I'm going to pass D type, bigger type. B, D type parameter. Is at this single course, I'm going to type INT eight. Then close the round races. This is the data type of an image. And if you want to see this image, just you need to type this command. You may show copy this command and hern pist and I'm going to replace at with ban. Also, I want to change the variable. I NG two blank, and I'm going to set this file. After set this file, I'm going to run the scole. So I'm going to click this Run button. After click this run button, as you can see, now you can see, now you trend two Window. One window is KAT image and another window is blank image. And as you can see, this is a squaring image, 500 by 500 pixel. So instead of using KAT Image, I'm going to draw my object in this blank image. So here, I'm going to comment out the lines, these two lines. Otherwise, we can remove it. For now, I'm going to comment out. Now at first, I'm going to paint this blank image in a certain color. So I'm going to type a coman paint the image in a certain color. To fill the blank image with new color, here you need to pass another. We need to pass attribute four color channel. We are going to use to three color channel RGB. Now we can able to use three color channel value. So to color the blank image, first time going to type blank, then you say the square ress I'm going to pass codon. It's me or the pixel I go to select all the pixels. Then then I want to fill it with green color. So I'm going to pass 0255, 255, zero. It's going to fill this black image with green color. Then we need to run image show function. So here to type dot Imho the runderF to provide a nameFirst we need to provide a name for the window and I'm going to type green. At the next value, we need to pass the variable. The does. And I'm going to set this. Up to set this file, if I run this command, you can see it with a two window. This is black window. This is the default one, and this is the new window with green field color. And to get different color, just you need to change the values. Suppose hand of pass zero comma zero, come up, 255. I'm going to set this file and run this code a. Up to run this code, now you can see a different result. This time, this window filled with red color. He we use slicing technique to select all the pixel. Not only that, if you want to fill a certain area with colour, yes, you can, you need to provide the vile of starting point and end point. Suppose at our first view I'm going to pass 200 by 300. At the next, I'm going to pass 300 go 400. If I set this file and run this code, now you can see it at a red square box at that position. So this is how we can fill recolored. The next thing I'm going to do, I'm going to do to draw a rectangle. So first, I'm going to remove both the lines because I don't need it, and then I'm going to type a single leg command. So I'm type hashtag and I pass draw a rectangle. To draw a rectangle, we are going to use a method, which is cv dot Rig tangle. This one. And to draw the rectangle first, we need to pass an image. So inside the round recess, I'm going to pass black. And also the function take multiple argument. Image 0.1 0.2, color, thickness, and if you want to provide line type, yes, you can. So for the starting point, inside the round ress I'm going to pass zero comma zero, and for the end point, I'm going to pass 250 pixel by 50 pixel. Then we need to pass color. And for the color value, I'm going to pass 02550. And here, also, we need to pass thickness, Sun type, thickness, and I'm going to pass thickness Vo two. And when it's done, also we need to show this image. So I'm going to type cv dot image show IM, so inside the roundresses first will to pass a window name. So I'm going to type rectangle. Comma. The pass the image. In which image, I want to apply the rectangle, which is black. This one. I'm going to set this file. After I set this file, if I run this code, you can see the result. Now you can see that window in this window, it draw rectangle. But you notice Harry create total three window because we call image show function three times. So for now, I'm going to comment out. So for now, I'm going to comment out this line. This one, also, I'm going to comment out this one. I don't need it. Also, I don't wonder for the square box, so I comment out it. So if I run this code again, you guys see the result. This is the rectangle we create in the blank image. And as you can see, we start this rectangle from the origin position. And if I change the origin position, suppose 50 by 50. Then set this file and run this code again, you guys see the result. Now it start from at that position. And if you want to increase the height of this rectangle and you want to use full of the image height, in that case, just pass. Again, I'm going to start 0-0, and this time I'm going to pass 500 by 500. And I'm going to set this file. After I set this file, I'm going to run this code again. Now you can see it cover the whole image frame with the rectangle. And if you want to use the hub of area, just you need to pass 250500 and set this file and delete this window. After set this file and run this code, you can see the result. So you can define size and position of this rectangle from this vets. Now you decide you want to fill this rectangle with color. For that, just want to change the thickness value. So at first, I'm going to delete this window, and then I'm going to change the thickness, thickness equal to v dot field. So after set this story if I un this code, you can see the result. As you can see, this time I want rectangle fill with green colored. Now I'm going to show you something. Instead of passing these values, you can use shape. You can use exactly bland shaped area. Let me show you. So I'm going to duplicate this line and comment on previous line. And this time, instead of using numeric value, I'm going to use shape. So I'm tied the image name blank. Blank, dot, shape. Then inside the square resis, here I'm going to pass one. And then I'm going to divide this value with two. Then I'm going to pass comma and I'm type blank dot shape. And inside this square resis I'm going to pass zero, and I'm going to divide it with two. If I set this slide and run this code, basically it's going to create a square shape because as you can see, here is at height, 500 and width 500. And if we divide 500 with two, then it's going to return 250. Also, if we divide this 500 with two, again, it's going to return 215. So it's going to create a square. Basically, we are going to scale down the shape. So after set this soil if I run this squad, here you can see it creates a square shape. And its height is half of the image height, and also is half of the image width. At the same way, if I divide it with three and then s up this file and run code, now you can see a different shape. This time, height is divided with three and image wide is divided with two. That's why it created a rectangle shape Basically, I want to show you instead of using exact numeric value, you can use the shape to calculate the value. Now moving on to the next shape. Let's draw a circle. So here I'm going to type a single line comment, and I'm type draw a circle. And also, I'm going to comment out the lines. I don't need it. To create a circle, we need to use cv dot circle method, cv dot circle. And inside the roundss first, we need to provide the image which is there. Then we need to pass the coordinate of the center position. So inside the round addresses, for now, I'm going to pass mid position of this image, which is 25250. Then in this function, you need to pass the radius wave. And here I'm going to pass radius value 40. Also, we need to provide the color. So inside the round recess, I'm going to pass 002 55, which is going to return red color, BGR. And then we need to provide the thickness of this border. So tight thickness, I need and I'm going to pass three pixel or three. And then we need to display this image. So Hemo type cv dot image show Im show. Then inside the round presses first want to pass the window name and window name is circle. And the next window I'm going to pass is the image name which is there. And I'm going to set this file. After I set this file, if I run this code in my terminal, this is what it create. It creates a circle at the middle position of this image. And now, if you want to fill this circle, then you need to change the thickness viel. But this time, I'm going to use another do. Instead of using Cv dot field, here I'm going to use minus one. If I pass this value and set this file and run this code again, now you can see our circle filled with red color. So this is another method that you can elmate with color. Just you need to pass thickness value minus one. Also, you can use cv dot fill it. And now I'm going to show you how we can draw a standalone line using open C. So I'm going to dig this window, and first, I'm going to comment out both the lines. And here, I'm going to try to comment, and I'm going to try draw Hey, line. To create the line, we need to use v dot line method, cv dot line. Then ins the round dress, first, we need to pass the image where are we to draw the line, which is black. And now we need to pass the coordinators of this line. For that, I'm going to use the same values. So I'm going to copy this section and paste up to that. Just I'm going to change the thickness value this line. Here I'm going to pass thickness value three. Also, I'm going to change the color, and I'm going to use white color, fully white. So I'm going to pass 255255, 255. After I set this file, if I run this code in my terminal, you can see the result. And before set the file, I run this code, we need to show this image. So I'm going to copy this section, paste it here and I'm going to rip the circle with line. That's, I set this file and run the code. After I run this code, you can see the result. So this is the light we create using open Ceri. And if you want to change the origin position, just need to change the view. Suppose I'm going to start it from 2020 do. If I set this file and delete this window and rewrite this code, you can see this time it start from at that position. So we already cover most of these shapes. The last thing I'm going to talk about text, how we can put text on image. For that, first, I'm going to delete this window, then I'm going to comment out the lines. Also, I'm going to type a comment. This time, we're going to write text on image. It is very straightforw art to put text on image. First, you have to type cb dot, and there you need to call put text method. Put text. Then inside the roundsesF we need to select the image where we are going to put the text. In our case, blank. Poma, ten, we need to provide the text, which you want to put in our image. In our case, hello n. Then we need to define the position from where we want to start or text. So inside the round phase, I'm going to pass 225 comma 225. Next, now we can specify font phase. So here to pass the font phase and to pass the font phase, we need to use some OpenCV in below font phases. And we are going to use cv dot font, underscore, here we have a lot of options. We have, complex small, duplex plane, but we're going to use x, this one. And the next will we need to pass for scale. And I want to scale up to 1.0. And next, we need to provide the font color and her pass zero, 255 comma zero, and also to pass the thickness. And for thickness, I'm going to pass something. I think two is enough, and oops, I need to provide command between these vs. That's it. Then we need to call image show function, the dot Imho. Then inside the roundre at first we need to provide a window name which is text. Then we need to provide the image black. And I'm going to set this file. After I set this file, if I run this code in my terminal, you can see the result. So as you can see, we successfully put taste on this image. So I hope it's clear for you how we can create different shapes and lines. Also, we can put taste on an image. So this is it for this tutorial. So thanks for watching this video. Sad for the next tutorial. 6. Important function in open CV: Hey, everyone, and welcome back to another video. In this tutorial, we're going to talk about basic function in open saving. Now, let's start with first function. Using this function, we can convert images into grayscale image. We already select this image and display this image. It's a BGiRimageT channel color image. BgiR stands for blue, green and red. Now there are ways that we can convert BGiRimage to grayscale image. For that, we are going to use CVD color method. First I'm going to declare a variable gray equal to CV, dot, CV decolor. There Is the round presses, first, we need to provide the image, which image I want to convert into grayscale, and our image variable is IMG Now at the next parameter, we need to pass the color code, which is CV dot, colored, underscoe BGR, to grade this one. Next, I'm going to display this image. I'm going to copy this function image show, and I'm going to paste it here. Then I'm going to replace cat gray image with gray variable. I want to set this file. Up set this file, if I run this code in my terminal, you can see the result. This is the color image and this is the grayscale image. It is very easy to convert VgRimage into grayscale image. At the next function, we are going to discuss how we can blur image. Here I'm going to type a single comment and I'm going to type blur. I'm going to move this line above the blur coma. There are multiple blur technique in Open CP, but we're going to use Gersionblur. At first, I'm going to declare a variable blur. Blur equal to Cv GusionblRtie, GH Gasianblur. Then inside the Runress first only to provide the image source which is IMG. Comma, next, in this function, we need to provide the kernel size value and we need to provide the value as a tapel here we mention the size of the blurred image. Is the round versus, I'm going to pass and here I'm going to pass kernel size three y three. Next, we need to define the border. I'm going to pass CV dot, border default. This one. Then we need to display this blur image. I'm going to use the same function, image show. I copy this function and I'm going to paste it here. First, I'm going to replace the window name, gray with blur. Then I'm going to replace the variable name, our image source, which is blur. I'm going to satisfy. But before I run this code, I'm going to comment out these two lines. I don't need it anymore. So after set this file, if I run this code in my terminal, you can see the result, but it is not too clear. In this image, we add little in it. If we compare it with the original image, let me show you I delete this terminal and commend this line. Also, this time I'm going to increace the kernel value. I'm going to make it three to 77 by seven and the set this file and run this code in my terminal. Now you can compare both the image. This is the normal image and this is the blur image. Here we add Gersionblur. Using kernel value, we can add more blood in English. Now, let's jump into the next function, which is age cascade. Here I'm going to create a single line coming and I'm going to type, age cascade. Using this function, basically we find edges in an image. There are many H cascade available in open C. But in this tutorial, we are going to use CIH cascade, which is pretty famous in computer vision world. Basically, it is a multi step process that involves blurring, grading, and many. So let's start. So first, I'm going to declare a variable, and our variable name is an Canny because we are going to use any cascade, assigned with C we dot Cy. Inside the roundresses at first, we need to provide this source image IMG W that also we need to pass to threshold value. First value I'm going to use 125, and the second value I'm going to use 175. Now, let's try to show this image in our window. But before I'm going to copy this function, Image show and comment out this function. Then I'm going to paste it here and replace BlaiKN Also, I'm going to replace the variable name, Ken. Now let's run the code in our terminal. After run this code in our terminal, this is what it looked like. This is what it written. So as you can see in this example, using as cascading, we are able to detak the edge of the SCAT Image. Not only that, also we can reduce the edges if we pass a blur image. Let me show you. So this time, I'm going to pass blurry age a Em source, this one. I'm going to replace ING with blood, and I'm going to set this file. After satis file, if I run this code in my terminal, now you can see, we are able to reduce the noise. Because of using blurry Image, that's why it reduced the noise. But if we reduce the blood value, let me show you if I use Gerson blood, three y three, and then this file, if I uncode, now you can notice the result. Now you can see it increases the edges. It increases the noise of this age. If you change the blood value of an image, then you can detect he more clearly. At the next function, we are going to discuss how we can dilate an image. To dilate the image, we are going to use specific structure and element. The structure and the elements we are going to use actually are the ages. At first, I'm going to delete my terminal and then I'm going to type a single comment and type dilating the mache. And then I'm going to declare a variable. He type dilate it equal to Cv dot, dilate. Then inside the roundreesF I'm going to pass the CE image, this one, a Image so. To pass the variable CE then we need to provide the kernel size. After inside the round dresses, I'm going to pass 33, and also it iterations. I'm type iteration one. Now let's display this image. For that, I'm going to use the onto function and I'm going to paste it here and replace Canny with dilate it. I copy this variable name and replace Canny with dilated and also we need to replace the variable. Then I'm going to set this file, and then I'm going to set this file. Let provide some space between the functions. Otherwise, it look very glancy. Then I'm going to comment out this line. I don't want to show the original image this time. I want to show only the canny image and dilated image. Please satisfy if I run this code in my terminal, you can see near the same result. This is the dilated image, and this is only the Ky image. These are the edges and these are the dilated edges. If you notice carefully, you can see the dilated edges are much more thicker than previous one. If I increase the iteration value, let me show you. I'm going to close the terminals. And then I'm going to increase the iteration Willow. Oh, I'm going to make it. So I'm going to make it three and then set this file. Then I'm going to run this code again in my terminal. After I run this code in my terminal, you can see the result. Dilated edges are much thicker than C edges. This is how its work. Now I'm going to discuss the lst function, which is how we can resize and crop image. Here, I'm going to type a single end command and I'm going to type resize. We already learned how we can resize our frame in our previous videos. This function is very easy to use. First, we need to declare a variable and our variable name is resize. Equal to Cv dot resides. Then inside the run recess, first, we need to pass the source image, which is ING. Here are the pass ING, and next, we need to pass the dimension of the image. I'm going to set it 500 by 500, but it's going to ignoring the aspec ratio. I'm going to comment out what the previous onto function. I don't need it for now and I'm going to set this file and also need to call the function p the resize image. To replace this variable with resins. Also, I'm going to change the window name resides and set this file. Now let's on the code and see what it written. After on this code, you can see the result. It rest our image successfully, but it do not follow the aspect ratio. If I compare it with the original image, let me show you. If I uncommen this line and the set this file again, and this time around to pass 400 by 400, 400 by 400 and set this file and run this code, you can see the result. This is the original image, and this is the resize image, you can see the difference between their aspect ratio. Next, I'm going to show you how we can cropping the image. I delete the windows and I'm going to tie another comment, cropping. On the base of their pixel value, we can select the portion of an image. So let's drop the image. So at first, I'm going to declare a variable crop Tikal two, I'm going to pass the source image IMG. Then we need to use square braces and we need to pass the points. First, I'm going to select 50 colon 200. And the next point is 200 to 400. Then I'm going to display this image using image function. So I copy this function and I'm going to paste it here and I'm going to replace the variable size with Cro Also change the window name crop I'm going to set this file. Before I set this file, we need to comment out this line. Also, we need to compare it with the original image. So after set this file, if I run this code in my terminal, you can see the result. This is the original image, and this is the cropped image. This is the crop image portion of this original image. So if I try to match with that, yes, we can match with that. This is how it slowed. So that's pretty much for this video. We just learned basic function in open C. At the next tutorial, we are going to learn about image transformation. So thanks for watching this video. Stay tuned for the next tutorial. 7. Open Cv Image Translation part1: Hello, and welcome to Python Open City course. In this six set, we are going to learn basic image transformation. These are common technique you would like to apply on your images, including translation, rotation, resizing, flipping. Let's start with translation. Translation is basically shipping the image along the xs and yx. Using translation, you can ship the image up down left right. Also you can mob it combinedly. To translate the image, we can create a translation function. Let's declare a function. For that, I'm going to tie BA and our function name is translate. Translate inside the Rundreses and I'm going to pass throughout three parameter. First, we need to pass the image, which image I want to translate IMG. Then I'm going to pass X xs and next I'm going to pass YXs. Then inside this function, now to translate the image, we need a translation matrix. So I'm going to declare a variable transmt. Transmat require two, and I'm going to use a function called float float 32 from the Numbi module, NP, tot, float, dot, float. 32. But before we need to import the NumPi module. Here we type, import NumPi as NP and this one, take a list and inside this list, have to pass to other list. Another list. In our first list, we're going to say one, zero comma X, and in our second list, I'm going to pass zero, one comma Y. Once we create the translation of the matrix, then we need to provide the Tmage. I'm going to declare another variable called tension. Dimension equal to am pass a tuple. Inside the round ress here Amul pass, image shaped methyl. Image shape inside the square ss, am type one, as you know, it is defined the image with. Also, we need to provide the image height. For that, we need to type image, dot, shape. Inside the square dresses, I'm going to pass zero. It's defined liquid. Then we can return some type, return vt fine. F fine. Inside the round presses, I'm going to pass the image. Then I'm going to pass the matrix transmt and we need to pass the dimension. Now let's create our first translated image. But before I'm going to tell you if I pass minus X value, then it's going to shift the image lip side, and if I use minus Y value, then it going to ship the image upside. And if I want to move the image right side, then we need to pass positive X value. And if I move it downside, then we need to use positive Y value. Now, let's create the translated image. So to type another variable, translate it. Translate it equal to translate. Translate, then inside the round process, we need to pass the image. First value, I'm going to pass this image, ING. Then we need to pass the xs direction and xs direction. I'm going to pass positive hundred value for xs for YXs also I'm going to pass positive hundred. After satisfy, now we need to display this image in our Window. For that, we need to use c dot Imho methor Cv dot Imo Image show. Then instead the roundreF we need to provide a name to this window, which is translated. I copy this variable name and I paste it here. Next, we need to use the image. Which image I want to fw. I want to fold this image translate it. I call this variable and I'm going to set this file. After set this file, if I run this code in my terminal, here you can see the result. That's the difference between two rezar. This is our original image, and this is our translated image. Here we ship the image, 100 point xs and positive 100 point yxs. It's shipped down by 100 pixel and shipped right by 100 pixel. Now, let's ship the image, left by 100 pixel and down by 100 pixel. This time, I'm going to use negative values. I want to move it -100 pixel Xs direction. I want to say lip side and also, I want to move it downside. So our X va in sale. Now, after set this file, let's run the code in our terminal. After run the code in our terminal, you can see the result. As you can see this time it moved the image lip side and also it moved the image downside. But if you want to move the image upside, in that case, just give it to pass minus YXS value. -100 pixel. Now let's run the code and see is it work properly or not? Up to run the code, you can see the result. This time we ship this image, 100 point left side and 100 point upside. This is how translate work. Also in our commence section, I mentioned the positive and negative value working method. If you want to move it at left side, then use minus X, and if you want to move it upside, then use minus Y. If you want to move it right side, then use positive X, and if you want to move it downside, then use positive Y. This is it for this tutorial at the next part we to talk about rotation. 8. Open Cv Image rotation part2: Good to see you back guys. In this section, we're going to learn how we can rotate image using Python OpenC. As you can see in my Visual Studio code editor, Harry already created Python file, rotate dot pi, and I already distend image using image it function. Then rotation is exactly what it sounds like. Rotating an image in some angles. At first, I'm going to create a function, Dave and our function name is rotate this function going to take total three argument, the image, and next it's going to take angle, the rotating angle. Also, you take another parameter which is rotating point, some t, ROT, point, equal to nun. Then inside this function, we need to grab the height and width of this image. For that, instant the rounds amo type height, width equal to, I'm going to use ING shape method, Ng. Inside the square Basis Hemotype colon two. I mean the first two value, which is height and. Next, I'm going to use IP condition. If the rotate point is none, then inside this IP condition, we are going to rotate the image around center and to get the center point, I'm going to type another variable root point to get the center point, her type rot point equal to inside the roundress, we need to divide with with two. So type width, divided, divided by two. Also at the same way, we need to divide height with height divided by two. Is going to return the center position, otherwise center coordinates of this image. Now we need to create the rotation matrix. For that, I'm going to tie first I'm going to declare a variable rot man rotation matrix. Rotation matrix equal to v dot get rotation matrix. Gate rotation matrix two d, this one. Then inside the roundresses, here you need to pass the rotation point, which is center position. I'm going to pass this value, rot point. Ot point. When that also we need to pass the angle. Then I'm going to pass scale vial. If you want to scale the image, when you rotate the image, in that case, you can pass to otherwise, otherwise, you can use the default scale value, which is 1.0. Then we need to mention the dimension as tuple. Type dimension dimension equal to is at the tuple, I'm going to pass width and height. Now we can return the rotate image, SmtiRturn here I'm going to call the method v dot, fine. Then inside the round brass, here first, I'm going to pass the image. Then I'm going to pass the rot mat, rotation matrix, and also to provide CA between them. Then we need to pass the dimension. So that's the function that can rotate the image. Now, we need to call this function. For that, I'm going to create a variable named rotate. Rotate equal to here I'm going to call the function rotate and inside the roundres first, I'm going to pass the source image IMG, then I would pass the rotating angle, which is 45. I'm going to set this file. Now we need to for this image in our window. For that, we need to use image function. So here type v dot Im inside the round press, at first, we need to provide a name to this window which is rotated. I copy the name and pass it here. Then we need to use comma and we need to pass the image source, which is rotated. I'm going to set this coin. After set this file, if I run this code in my terminal, you can see the result. You can see this is our original image, and this is our rotated image. Here we rotate the image counter clockwise. If you want to rotate this image clockwise, just you need to pass negative video -45 degrees. After satisfied if I re run this code, you can see the result. This time it rotate the image clockwise. This is how we can rotate an image using open saving. We can rotate this image at any angle. This is it for this tutorial. Thanks for watching this video. 9. Open Cv Image Translation croping part 3: So welcome to another tutorial related transformation. In this tutorial, we are going to learn resizing, flicking and cropping. So let's start the practical. We already learn about resizing in our previous tutorials, we need to use Cv dot resize function. So at first, I'm going to declare a variable named reside. Resized equal to C dot resize function, resides. Then inside the rounds, first, I'm going to use a variable IMG image inside the rounds, we need to provide the resize dimension. I have to pass 300 by 300. Next, we need to pass the interpoliion value. I have typed interpolation and there, I'm going to use v dot InterqPT one. Enter Quvit is slow, but it return better image. Then just agree to call the image show function and place it here, and this time, at first, I'm going to change the window name. This is resized. Also, I'm going to call the variable, this one. I'm going to set this one. Now, let's set the file and run the code in our terminal. After run this code, you can see the result. This is our original image, and this is our resized image. Now let's talk about our next transformation, which is flipping. First, I'm going to comment out this line. Then here I'm going to learn how we can flip an image using open sNing. For that, we don't need to define any function. First, I'm going to type a single length command and I'm going to type flipping. Then I'm going to declare a variable called flip. Slip equal to c dot flat. Then in the round presses, I would pass the image which I want to flip IMG. Here we need to pass a u. It would be zero, one, otherwise, negative one. Let's start with zero. Then I'm going to display this flee. For that, we need to use image show function, and I'm going to replace the window name, resize with flip. Also, I'm going to replace the variable laid. I'm going to set this file. Zero basically implies flipping the image vertically, and if you pass one, then it's going to flip the image horizontally. If you pass minus one, then it's going to flip the image, both the direction, horizontally and vertically. Let's start with vertically, pass zero. After set this file, if I run this code in my terminal, you can see the result. Vertically it flip the image. Now I'm going to lit this image horizontally. I pass one. After set this file, if I run this code, now you can see it flip the image horizontally. Basically it mirroring the image. You can see the mirror reset. And if you want to flip this image both the direction for that, you need to pass minus one. If I set this file and run this code, you can see the result. Now you can see it flip the image, both the direction, vertically and horizontally. This is how flip work. Next, I'm going to comment out this line and this time, we are going to learn about cropping, how we can crop an image. We already learn about cropping in our previous tutorials. First, I'm going to declare a comment and Healt cropping. Then I'm going to declare a variable crop Crop equal to then we need to pass this source imagemG. Then you side the square recess, we need to define the position, the point, the starting point and the ending point of the crop. Here I'm going to pass little slicing technique 200 cold 400. This is the starting point, and for the end point, I'm going to pass 300 400. This is the point, and I'm going to satisfyle. After set this file, we need to display this image. For that, we need to use image show function. So I copy this function and paste it here and replace flip with ra. Also we need to replace the variable length crop and we need to set this file. After set this file, if I run this code in my terminal, you can see the result. Oops, it's not working. I think I did some mistak. Whoops, how you used a comma. We need to use colon Si. After then again, I'm going to set this file and run this code. If I run this code, you can see the result. This is what it return. This is the cropped image. I think this is this exact position. If I change the crop value something 50, by 200 and 100 by 200. Then set this file and run this code. Let's see. This is the area it returned. We already learn about that in our previous tutorials. So this is it for this tutorial. In the next tutorial, we are going to learn contre detection. Thanks for watching this video, Styune for the next tutorial. 10. Contour Detection Opencv: Hello, everyone. Welcome to another video. In this video, we are going to talk about how we can identify contours. Cues are basically boundary of objects. It made with line and curves. In a simple sentence, contour is the outline or shape of something. It is especially a curve, otherwise, irregular figure. The contour word came from a Italian word, which is contour. But if I talk about mathematical point of view, Ctures and edges are different thing. Oturor very useful thing when you involve in shape analysis. Also it useful for detection and recognization. In this tutorial, I would like to give you the idea of contours and how we can use it in open C. As you can see, her we read a image file and this image using IMF function. If you learn this code in my terminal, you can see this image. It is our CAT image. It is the exact CAT image. The next thing I'm going to do, I'm going to convert this image into grayscale image. For that, hemotypeGray. Gray equal to CV dot. I'm going to use a method called CVD color. C VT color inside the round resssF we need to pass the source image, which is IMG. So to pass this variable. Then I'm going to use cv dot color. Cv colored. Color and I want to convert this color BGR green. Underscore BGR gray. This one. And then we need to display this image. I'm going to copy this line and paste it here. I'm going to replace Tati gray. Also, we need to replace the variable name, gray. I want to set this file. After set this file, if I done this code in my terminal, you can see the result, you can see the different. This is our an image, and this is our grayscale image. After this, I want to grab the edges of this image. For that, I'm going to use E edge detector. At first, I'm going to declare a variable KE equal to C dot K. Then inside the round versus, at first, I'm going to pass the source image, which is ING. Then then you need to pass total two threshold. And here I'm going to pass 125 comma and the second value is 175. Then I'm going to display this image. I'm going to copy this function I am show and paste it here and replace gray with K. Also, I'm going to comment out this actual colored image. I don't want to show three image in my display. Upset this file, if I run this code in my terminal, here you can see the result. This is the grayscale image, and this is the edge detecting image. And now I want to find the contours in this image. So we are going to use contour method. Let me show you. So here I'm going to use this method and this method to basically written contours and hierarchies. So here I'm going to type contours the second variable is hierarchies. So here, I'm going to use the method Civ dot find contu. Then inside the round ress I'm going to par the edges as can. They take moves in which mode it find the contues. It could be civtRorcT. Let me show you Civ dot retorscoeTT one. There are a lot of tour. We have external mode, if I show you rear, external com list, floodF now, I would like to go with list because it's going to return all the contour, without using tree, I would like to go with list. The next method I'm going to pass, which is contour approximation method. So after COVA, I'm going to type Cividt chain approach, nun, this one. Now, let's explain this line again. Here you can see CV fine contues. This function ret condues from a binary image, and Ki is the source image. In which image you want to detect the conduce. CVR least is the mode of contu retraable and the next method, CV dot chain approximant I store all the point of the contues and hierarchy is the variable. It contained information about image tropolos and the contour variable is the list of all contues. It's going to store all the contours which we find in this image and each contour store a Nampi at a point. As I told you, contour is a list, and if you want to print how many contour we find in this image, yes, you can. Just you need to use print function print then inside the round recess, you need to use len function A ten. Then inside the round recess, I'm going to pass the contours. This one, and I'm going to set this file. After set this file, if I run this code in my ravinal, this is the result we found, which is basically it's returned 359 contoures. And if you want to reduce the noise of the contoures, you can blur this image. So let's blur the gray image. So here, I'm going to create another variable blur. Blur, sorry, not blue. Blur, equal to v dot Gaussian blur. Then inside the round recess, we need to pass the source image which is gray. With that, we need to pass to values, how much gray I want. After pass the source image, also, we need to define the K size. I want to say kernel size. Then inside the round resis, we need to pass the kernel size. So here I pass five CA five. Then we need to pass another value, border time. So I'm going to type CV dot, border, and here I'm use the default one, border deform. Also, I would like to display this blur image. So I copy this function, I am show and I'm going to paste it here and replace gray with blur. Then I'm going to set this file. After set this file, if I run this code in my terminal, you guys see the result. This is the blur gray image, and this is the Kyimage. Next, I'm going to find the edges of this image. For that, we need to pass the variable and that section. Here, can. I want to satisfy. Appro satisfied. If I run this code in my terminal, this is what it return. Now you can see it return total 23 contu. It reduced the number of contour because hero use lnge. This is how we can find contu from an image. There is another way that we can find contour from an image. But I'm not going to cover in this trutorial. I'm going to cover it in the next tutorial. 11. Contour Detection Opencv part 2: Hey, good to see you back, guys. This is the second part of this tutorial. In this section, we're going to see how we can use different method to find conduits. Without using Canny edge detector, we are going to use threshold. It is another function from OpenCV. So at first, I'm going to commit out on the lens. I don't need it. Then down there, first, I'm going to declare a variable, with. Also, I'm going to declare trash. Equal to cv dot h. This one. And it's going to take this gray image. So I'm going to copy this in gray. Also, we need to pass threshold value. The first threshold value I'm going to pass 125, and the second threshold value I'm going to pass 255. Don't worry about it. We are going to learn about threshold in our upcoming tutorials. Threshold basically binarize the image. If the intensity of A pixel is below the 125, then it's going to set to zero. I want to say, then it's going to set to black color. And if the image intensity is above the 125, then it going to set then it's going to set to 255, otherwise, white color. And in our fine contour method, we are going to pass this thrash variable. Trash. Not only that, here we need to pass another value in threshold function, which is threshold t. So the pass comma, and here I want to tie cv dots BinarT one. Now, set this file and run the code in a terminal and see how many contour it return. Up to run the code, as you can see, totally return 1073 contoure. Now, let's display the thrashimage, not the gray image. So I copy this line and comment out this line and call the onshoFunction, and replace gray with trach. Again, I'm going to set this file and run the code in my terminal. Up to run the code in my terminal, this is the result we farm. This is the result it return. This is the Black and white image. No, it is the binary image, and it's made with black and white dots. Don't worry about it. We are going to talk about in our advanced section. Just one thing you need to remember, hresold basically binarize the image. It is the binary form of an image. Zero indicates that color and 255 indicates white color. Now, let me show you one thing. At first, I'm going to import num pi. He tie, import num Pi as NP. Apart from these, I'm going to create a blank variable B Blan equal to, and here I'm going to call function from NumPiNp dot Os. Then inside the round recess, I'm going to grab the image shape, this variable shape. So here, I'm going to type mg dot Shae. Then inside the square resist colon two, I'm going to grab all of this to v. Next, we need to pass the data type B, type, and here inside this single quote, I'm going to pass UI NT, H. And then I'm going to display this bank image using image show function. I copy this line and paste it here. And rip this image with blank. I'm going to set this file. Up set this file. If I run this code in my terminal, you can see the result. This is the blank image we create. This is the exact dimension of this CAT image. Now the question is, for what purpose we create this blank image? Because in this blank image, I'm going to draw or contue. I'm going to create this contute in this blank image. That's why I create this blank image. And to draw the contours, we need to use a function called draw contube so type v dot, draw contes this one, and inside the round dress, first, we need to pass the blank image. So type black. Next, we need to pass the contour list. This variable. As I told you, this contour is a list. I store all the control ws. So to pass the contra contuses. Next, we need to pass contour index. I want to say how many conti you want to pass, and I want all Op. So I'm going to pass Native one. Next, we need to provide the color, and we need to provide the color in Bzier format. So I want red color. 00 comma 255. Also we need to pass the thickness and I want to pass thickness two. And now we need to dist this blank image. So to copy this line, I am sure and I'm going to paste it here. And also, I comment out the previous line. I don't need it now. And I'm going to set this file. After I set this file, let's run the code in our terminal. A I run the code in my terminal not working properly. I think there is a mistake in my code. I think I did some mistakes, so let's check it. Here, this one, I don't need this slicing technique. Just want to use image dot shape. It's going to grab the image shape area. So let's set the file and rerun it. Up to run this code, as you can see, this is what it returned. It returned the cat image. Basically, here we draw a new image using the condut points. First, we take a blank image, then we use the condued points from the thresh word image and apply it in the blank image. It is not the same thing. From the threshold image, go to grab the edges and using these edges, we draw an image in our blank page. If I reduce the thickness, I think it looks more prettier. Let's see. I'm going to reduce the thickness 2-1 and re run this cool. Yes, nerves look pretty clear. Also, you can get the different result if you change the threshold value. Suppose if I reduce the threshold value 255-175, then it's going to return a different result. Don't worry. We are going to learn more about thresholding in our upcoming tutorial. So thanks for watching this video, Statute for the next tutorialm. In the next tutorial, we are going to talk about color spaces. 12. Color Spaces: Hello, guys, good to see you back. Once again, I am back with another tutorial related open CV. And in this tutorial payer we will learn color spaces. Color spaces are basically space of colored. RGB is a color space, gray scale is a color space. We have also different color spaces such as HSV, AAD and Mini. Now, let's start with to try to convert the image with gray scale. So we are going to convert GR image to grayscale image. As you know, Bier is the default color mode in open C. For that, I'm going to declare a variable gray, gray equal to assigned, CVI dot, I'm going to use a method called CVD color, CVT color. Then inside the rounds, first, we need to pass the source image which is IMG, then we need to pass the color. Smit color. Underscore, and I want to convert BGR to gray, this one, and I'm going to sub this file. Then I'm going to display this image using Imho method. Here tight, Cm dot, IMO. Inside the round brcess first, I'm going to pass the window name which is gray. Next, we need to pass the source image, which is gray variable. And I want to set this file. After set this file, if I run this code in my terminal, as you can see, this is our original image and this is our grayscale image. Here we convert BGR image into grayscale image. Now I'm going to convert BGR to HSV format. HSV stands for hue saturation W. First, I'm going to type a coman BGR HsVT apply the HSV color and want to copy the same codo. And based it here. First, I'm going to change the variable name grade two HsV. Then I'm going to change Bijiac two HSV. This one. Also, I'm going to change the window name, replace gray with HSV and variable length. I want to satisfy. After satisfy, if I ra this code in my terminal, let's see what it return. This is the gsV color mode of this image. Now, let's talk about another type of color space, which is AAB. Here I'm going to type another single command, hashtag. This time, I'm going to convert VGR to LAB. So again, I'm going to copy this section and I'm going to paste it here, and this time, I'm going to change HSB to AB. Also, I'm going to change the color code, jia to LAB. Then we need to change the window name, LAB, also in to change the variable length. After set this file, if I run this code in my terminal, let's see what it return. This is the LAV color mode of this image. This is how it slow. As you know, by default, OpenCV use gR format. But outside of OpenC we use RGB format, which is opposite of BGR. It's a inverse color mode of BGR format. Now I want to open this image in RGB mode. I'm going to open this image outside of the open C. First, I'm going to comment out on the lines, then I'm going to import Moni mat leave, some type, import, Mt leave. This one, He call a particular function Pi plot a plot. If you don't install this module in your system, then just open your Windows Power shin and tie PIP install Matt Blot and Tick Enter. It's going to install it in your system. I already installed it, so I'm not going to install it E. So mad Blot leap, read the image in AGM. But have we opened this image using open C. I read the image in BGM. Lets open this image using Mt Blot leap. He type PLT, dot, I would run IMS function, I am so. Then inside the runres I'm going to pass the source image IMG, image, this one. Then I'm going to display this image using PLT dot SFunction. Let's certify and runs code in our terminal. Appear on this code in our terminal, this is the BGiR mode of this image and this is the RGB mode of this image. Basically, it apply opposite color of this original image. This is BGiRimage and this is RGB image. Open read this image as Bzier mode and by default Mtlot so the image in RGB mode. That's why Matt Plotly thinks is an RGB color. That's why we can see the opposite version of this color. Now, let me show you something. I'm going to delete this window and comment out both the lines. Let me show you something. I comment out both this line and this time, I'm going to convert BGA to RGB colored. Here type of comment BGR two RGB. So first, I'm going to declare a variable RGB equal to CV dot CVD color, CVT color. Then is that the round res is image IMG, comma vt color. Underscore. This time I want to convert BGR to RGB. This one. Then I'm going to display this image. Hemotype CV that I am show. Is the roundreses first, we need to pass the display name, which is RGB, then we need to pass the variable name RGB. That's it. Now, let's run the code and see what it returns. Up run this code, also, you can see it written a RGB version of this image. It is now it looks pretty similar to matt because here we convert VGA to RGB image. But if I run this to code, copy this to code and paste it up that and comment out and uncomment both the line, and this time as variable, I'm going to pass RGB not image and set this file and run this code again. Now you can notice this time metap return the actual original image, which is similar with cat window. This is our original BGR image. Now, our RGB image and BGR image look same because we convert BGR image into RGB image using OpenCeM. Then I pass this image in image to function using matplot leave. That's why it returned this result. Just one thing you need to remember met plot leave default color code is RGB and OpenCV default color code is GR. So now you can see Harry convert Bzierimage to greyscale, Bzierimage to HSV, Bier image to LAB. But there is a downside on Open C. You cannot convert grayscale image into HSV. Also you cannot convert HSV to LAB directly. If you want to do that, first, you need to convert disk image into BGR, then you can change BGR image into HSV. I'm going to convert HSV to BGR image. Here type is single and command and I type this time I'm going to convert H a to BGR. And first, I'm going to declare a variable HHV underscore, GR. These are variable and this time I'm going to run C dot, CVT color. Inset this function, I'm going to pass at first, I'm going to pass the source image, which is HSV. I'm going to uncomment this section. And this time I'm going to pass HSV, this one. The next, I'm going to pass the color, which is CV dot, color, underscore and this time I'm going to convert Hs V to BGR, this one, I'm going to display this image. So type cv dot No inside the run verses, inside the single codes, first I'm going to provide a name which is HSV to BGR. Then I'm going to provide the variable. Before I set this file I run this code, I'm going to comment out all the image functions. I want to show the original one. Also, I don't know what to show these ones. I also found out these two lines. Uptrastep this file, let's sun the code and see what it returned. Up step this file, you can see the result. This is our original image, and this is our HSV two videoing. For better understand, let me show you the HSV image. If I show the HSE image, and I'm going to show this file. After I run this code, as you can see, this is the original cat image, I don't need it. This is the HSV version of this color, and we convert HSV image into BjerimageEgy. But we cannot convert HSV image directly into grayscale image. So this is it for this tutorial. In this tutorial, I tried to cover some little color spaces concept in OpenCV. So thanks for watching this video Stune for the next tutorial. 13. Color Channels: Hello, guys. It's good to see you back. Once again, I'm back with another tutorial real tate O pensimi. And in this tutorial, we are going to learn color channels. In this tutorial, especially we are going to learn how we can spit color channels, and also we are going to learn how we can merge color channels. A color image basically consists multiple channels, red, blue and green. All the images you see in a digital media made with three colored channel, red, blue and green. So image are made with the three colored channel, and OpenCV allow us to spit the colored channels. As you can see in my current working directory, we have an image, PIC dot JPG. If we open this image, as you can see, we have to do three subtle, three surgen with three different colors, red colored, green colored, blue colored. This is a single image, and OpenCV can fit the three colored channels, blue, green and red. So that's we are going to do in this video. Besidally we are going to use this image and split this image into three color channel. So for the three color channel, I'm going to take three variables, B, G R equal to, I'm going to speed the color channels. For that, I'm going to type Cv dot, I'm going to run spit function, Cv dot, spit function, speed. And inside the round press, I'm going to pass the image IMG. This split function going to split the image into blue, green and red color. But at first, quits convert this image into the grayscale image. And if you want to show this image in a window for that, I'm going to use, we can use Image show function, so type cv dot IM show. Then the round presses first, we need to pass a window name, at first I want to show the blue color, so I want to type Window name blue. C, then I'm going to pass the variable length B. Then I'm going to begin this line two t. Next, I want to show the green color, and for the green color, I'm going to pass G variable, and last one is for red color for the red color, I'm going to pass R variable. I'm going to stifle. Before I run this quote, let me tell you something. Basically is this function going to convert this image into Griskal image. Then according to the color channel value, we can see the dark and bright part. Let me show you. After set this file, if I run this code in my terminal, here you can see three razor. Now it's look pretty confusing. This is the original image. As you can see in the original image, we can see three of the color red, green, and blue. But if I show you the blue window, here you can see now blue color become white color. Now you can notice the blue window and the blue circles become white. Now you mi have cooten why it's become white, and other two colors become lack. As I told you when Opensi found the blue color at that image, then it increased the intensity of this color. And as I told you, si function convert the image into grayscale image. And whenever we run blue color channel in this gray scale image, then it highnt and represent the blue color with white color. And as you can see, this is the exact blue color. That's why it become white colored, exact white color. OpensNi did not found any blue color on this image. I do not found any blue color at green color circle. Also you do not found the blue color at red color circle. That's why it become black. At the same way, when we run the red color channel, this time, red color become white colored. As you can see, now red circle turn into white circle because this time it hich the red color and increase the intensity of this circle. At the same way, green colored channel work. This time it increase the green color intensity. You can do the same experiment with any colorful image. But for your convenience and to understand how it's work, we use this circle, three different color circle, red, green, and blue. And then groove circle, there is no green color available. That's why it become black colored. If any of green color percentage is available at that portion, then it become a little lighter. And remember, the original image come with three colored channel and the green window radio window and blue window come with zero color channel. Let me prove you that. If I delete this window and print the shape, let me show you type print inside the round ss, I'm going to type image, image variable shape. I'm going to get this line. And this time, I want to print one of the color channel low, suppose grey. I tie G. If I set this file and run this code to A, now you can see in my terminal, first we print the dimension 300 by 500 and then we print the color channel. In our first image, I want to say the original image, we have to draw three color channel. But in the next image, which is green window, we don't have any color channel. By default, we can call it one color channel value, but basically, black and white are not colors. Basically, in this window, black and white represent the particular color intensity. And now I'm going to comment out both the lines, and I'm going to merge this three image. But before I'm going to comment out this line, the original image. I don't want to this window, and to merge the three curve channel image, I'm going to use merge function. First, I'm going to declare a variable and our variable name is merge. Merge equal to Is the rounds Hem to coll a function called Marge. Sometime v dot merge. Is the round brass inside the square dresses I'm going to pass. Herem to pass list of three con channel, B, G R. And then I'm going to display this variable. For that, again, we need to run civet IMF function. In the roundress is first going to pass the window name which is merged. Copy the variable. This is the merged window and I'm going to pass the coma, and I'm going to pass the variable in merged. And I'm going to set this file. After I set this file, at first, it got for this three color channel window. Then it got for the merged window. Let me show you. So after run the code, you can see, first it print. At first, it displayed the blue colored channel, then red, then green. And at last, it display the merged color channel. And after merged the three colored channel, again, it return a colorful image. So this is for this tutorial. In the next tutorial, we're going to learn about blurring. So thanks for watching this video stadium for the next tutorial. 14. Blurring: Hello, guys, good to see you back. Once again, I'm back with another tutorial related Open C. And in this tutorial, we are going to learn blood. We are going to learn different method of blurring. Generally, we smooth and blur the image when we have noise in our image, and applying some blur, we can reduce the noise. In our previous tutorials, we already learn about gaussian blur, which is the most popular method for blurring. Without Gersion blur, there are many blurring techniques, and we are going to learn all of it. But before we start our practical, we need to understand something. We need to understand what is blurry. How actually blurring work. Here I'm going to open an image. This is our image. What exactly goes on when we apply blur. At first, we need to define something called kernel. Otherwise, you can call it indo. So here at that position, I'm going to draw a window. Suppose this is our window size. Here we draw a window over the image at a specific position. As you can see, this window has a size, and it is known as kernel size. And kernel size basically number of rows and number of columns. Here I'm going to draw three row and also I'm going to draw three columns. Cerdles made with row and columns or pixels. These are all pixels. So here we have total three column and three row, so its kernel size is three by three. Essentially, what happened? We have multiple method to apply blur. Blurred applied to this middle pixel, and this pixel is applied to result fraud rounded pixels. It apply the surrounding pixel result at the middle one. Let's discuss the first method of blurring, which is averaging. Let's get to the users to your code. And here, I'm going to type a single line command hasta and I'm going to type average. Now the question is how averaging blur is work. If I open this image again, as I told you, b always apply to the middle pixel and averaging blur work according to the surrounding pixel average. Suppose this pixel intensity value is two. This one is five. This one is seven. This one is three. This one is four. This one is two again. This one is eight, and this one is nine. As you, middle pixel intensity is four. Now, it's going to extract all the pixel intensity value and apply this value and apply this value to the middle pixel. This is how averaging work. Basically you want to say surrounding pixel average going to be applied to the middle pix. Let's back the users to your code. To apply the blur, first, I'm going to declare a variable and our variable is average. Average equal to CV dot Blur method. Then instead the run pass first, we need to pass the source, the source image, which is ING. And then we need to define the kernel size. And for kernel size, I'm going to take three column and three row asi. And we can display this image in our window, so I'm going to run CV dot If Mthood inside the roundre inside the single code, first, I'm going to provide a window name which is average, comma, and then I'm going to pass the variable, which is also avg. If I sat this file and run this code in my terminal, as you can see, it both the image. This is the original cat image, and this is the average blurred image. If I open it side by side, you can see the little difference. It is look pretty sharp, but this one is not sharp because here we apply average blur. And if you want to increase the blur intensity, then just to pass higher kernel value. Let me show. So without using three by three kernel size, I'm here to pass seven by seven and set this file and run this code again. Up to run this code, you can understand the difference. Up to increase the kernel value, it also increase the blurriness of this image. Now let's apply another blur, which is Gasianblur. Delete both the windows and here type another single comment Gersionblur. To apply the Gersionblur, herb to declare another variable guess equal to Cm dot Gersion. Is the round presses, first we to pass the source image, which is ING then we need to provide the kernel size and here I'm going to apply seven Casern. Also to pass Sigma X. It is used for X dictation deviation. For now, I'm going to set in zero, and I'm going to show this image on a window so I'm going to copy the on from Mthod and past it here. I'm going to rename Average with Gertian. Also change the variable name GUS. If I set this file and run this code in my terminal, now you can compare both the blood. This is average blur and this is Gasienbld. But you can notice here we use same kernel size seven by seven. But in average, it applied more blood, but in Gasian it look natural blood. That's why most of the time we use Gersienbl because it looks more natural. Then we are going to apply another blood. Let me show. Or next name is median blood. Hemotype is single and command median blood. Then Hembo declare a median. Median equal to C dot median blood. Then you set the rounds first L to pass the source image which is IMG. Instead of pass full kernel size, here to pass one intser. Suppose I'm going to pass three. Now open C we can assume that its kernel size is three by three. Now, let's display the image in a window, so I'm going to copy the Imro map hood and I'm going to paste it here. First, I'm going to replace the window name. I'm going to make it median. Also, I'm going to replace the variable length. Medium, set this file. After set this file, if I run this code in my terminal, now you can compare both the blur. This is the normal image. This is the median blur, and this is the Gasionblur and this is the average blur. Maybe you cannot understand the blurriness if I increase the value, let me show you. I'm going to did all the window and increase kernel size value. I'm going to make it nine and then set this file again and run this code again. Now you can see the median blur effect. If you compare the blur effect with other blur effects, it look like Oil t. Now, let's apply the st blur effect, which is bilateral blurring. So here, I'm going to create a single and comment and I will type. Bilateral blur. Bilateral blur is not a blur. It is a filter. It is highly effective in noise removal while keeping edges sharp. But this operation is slower than if I compare to other filters. Using bilateral blur, you can your image. Also you can maintain the sharpness of edges. Here, I'm going to declare a variable called bilateral, so I copy this name and I'm going to declare a variable bilateral equal to Here called the bilateral filter. Some type CV dot bilateral fitter. Then you say the rundresssF we need to pass the source image which is IMG variable, then we need to pass the dilateral in and I'm going to set the value five. Next, we need to pass this Sigma color. I'm going to set it 154 now. The next value, we need to pass Sigma space. Also I'm going to pass it and also I want to set it 15. Now, let's display the image. I'm going to copy the iron from Math wood and I'm going to paste it here. First, I'm going to replace the Window name, Blteral then I'm going to paste the very well name, Blteral I'm going to set this file. After set this file, if I on this code in my terminal, here you can see DE result. This is bilteralblA. I think it is pretty similar to our genome. Let's increase the value and make it nine. And also, I'm going to tell the Sigma values 75 and 75. After set this file, if I have the scorde in my terminal, now let's see what it written. Now you can compare it with the original image. This is how bilateral filter work. This is the original image, and this is the bilateral filter blur. So these are the four type of blurs we can apply in OpenCV. I hope now it's clear for you how we can apply different blur in open CV. So thanks for watching this video Studio for the next tutorial. 15. BITWISE operations: Hello, guys. Good to see you back. Once again, I'm back with a new tutorial related OpenCV. And in this tutorial, we're going to learn Batis operations. In OpenCV, there are four Biti operator. First one is N XR and the last operator is naught. In OpenCV, Bitois operator are used for pixel level operation between two image or between an image and a mask. These operations are useful for task like blending, masking and manipulation specific part of an image, and many more. As I told you, we have to tell for BTI operator. First, I'm going to type a single lend command here. No, I'm going to use multi length comen for that. Then in the multiline command, I'm going to paste all the operator's name and what purpose we use for. As I told you, this is the first operator, BTI. It applies a logical operation to each pixel of two images or an image and a mask. Next repertor we have BTs or it applies a logical operation. Then we have bitwis ZR. It applies a logical ZR operation last word is bitwise no. Basically, it invert the pixels. You can compare it with logical node operator. Remember, bitwise operator working in a binary mender. If we pass zero value, then it's going to turn of the pixel and if we pass one value, then it will turn on the pixel. I represent it white and black color. Now let's import another module, num pipe, Hemotpimport, numb Pi as NP. Then I'm going to take a variable called Blend and I would set it NP dot zero is at the RunssF, we need to provide the shape of this window. I'm going to take here I'm going to set window size 400 400. Also, we need to provide the data type. D type equals two is at a single codes INT eight. Basically, her create a blank black image and we say height 400 and width 400. Basically, we are going to use this blank image as basis to draw rectangle and circle. Next, I'm going to create a rectangle in this blank image. I'm going to type a variable RECt assigned with cv dot rectangle Method. Rectangle. Then inside the round ress first, we need to pass the image in which image we will used to create the rectangle. I want to use blank dot copy. I'm going to take copy version of this blank image. Then we need to define the margin. We need to define the margin from Xs and Y xs. I'm going to take it 30 30. Then we need to define the size of the rectangle. I'm going to take it 370 pixel by 370 pixel, then we need to provide color. As you know, it's a binary image, so I'm not going to use three color channels, so I'm going to provide 255, just white color. And at last, we need to pass the thickness. So I'm going to pass negative minus one because it will see this image with white color. After I create the rectangle, we need to create another circle. So I'm going to create circle variable, circle equal to CV, circle metal. Then inside the roundresses, first, we need to define the image. In which image I want to create this su type bland, dot, copy, then then we need to define the center position of this circle. I'm going to pass 200 by 200. Ma, then I'm going to provide radius to this circle, which is 200 and also to provide the color so I'm going to use single colored, so I want to pass 255, fully white colored. After that, we need to pass the thickness of this circle, so I'm going to pass minus one to fill the circle with white colored. Then we need to display both the geometrical shape in different windows. He run, C, dot, IN so method, No inside the round br inside the single first, I want to display the rectangle, so tie rectangle. Then we need to provide the variable name, which is rec. Next, I want to display the circle. I'm going to replace rectangle with circle and also we need to replace the variable length circle. I'm going to set this file. Now, after set the file, let's run the code and see what it will return. After I run this code in my terminal, you can see, here we have the two image that I'm going to work with. It is a rectangle and it's a circle. Now let's start with the first basic Btis operator. So let's start with our first Bitwise operator, which is some creative single and common Bitwise A. Before we discuss about bitwis and operator, let me show you what it does. So at first, I'm going to declare a variable bitwise a Bowie, underscore, N N, and equal to C dot, bitwise and operator. Bois and then then inside the round presses, need to pass to source. So as source, I'm going to pass the rectangle DC and as source too, we are going to pass circle. Oops, circle spelling is wrong, so I'm going to make it correct CR. And now I'm going to show this result in a different Window. Here I'm going to run CV, dot, on for method, M show. Then inside the round presses, first, we to pass the window name, which is Bitwise A. I copy the variable name as it is and paste it here. Then also to provide the source image, which is bitwise N, this variable. I want to set this file. After set this file, I'm going to run the code in my terminal. A unicode, as you can see, this is Bitois and operator result. Bitwise operator Bitwise and operator took this two image and placed it top each other and return the intersect reason. If I move this circle above the rectangle, then return the common intersect area of the combined shape, and it represents the common reason area with white colour and remove the other section. Now, let's jump into the bitwise or operator. Here I'm going to comment out this slide. Then I'm going to type another single and command and I'm going to type bitwise or operator or basically, bitwis orbor return the intersecting area it return the non intersecting area. Let me show you that. Again, I'm going to type a variable bitwise or Bitwis equal to c dot bitwis or operator. Then it's at the roundres I'm going to pass the rectangle. Then as source to, I'm going to pass the circle. If I display this image in a window, so I'm type C dot and informo. Then at the roundressF we provide a window name which is BTI R. Then we need to provide the source, the source image. After I set this file, if I don't code in my terminal, now you can notice, as I told you, it returned the intersection area, also it return non intersection area. It combined the common result with non common reason and create a new shape. Now moving on to the next Bitwie operator. For that, I'm going to write another single end command, but before I'm going to comment out this line and here I'm going to create another single endm bitwise. I'm going to type bitwise undersco Z operator, Bitwise and this operator name is O Zo. Let's see what it going to do. I create another variable, BitwisUnderscoe, Zo equal to. Basically, it's going to return the non intersected reason. Let me show you Sol type, CV, dot, Bis, Zr. Is the Runbss at first, I'm going to pass the rectangle. Then I'm going to pass the circle. And I'm going to display this image. So type, CV dot, I am fmthod. I am show. Then instead the undress, first, I'm going to provide a window name as window name and pass the same variable name, BTI zone. Then we need to pass the variable BTI zone. Now, let's set the file and run this code in our terminal and see what it return. Up to this code in my terminal, as you can see, as I told you, it's going to return the non intersected reason. If we put this image top of each other, it's going to create intersecting reason and non intersecting reason and Za return the non intersecting reason. These operators are very useful for STI. Now, let's talk about the last Bitwis operator, which is bitwisnO. So here, I'm going to type another single end comment and I'm going to type bitwisPtisn. Basically, it does not return anything, it inverts the binary color. It turned white into black and black into white. Let me prove you that. So here, I'm going to declare a variable called beat is underscope naught. Bitwise naught equal to Civ thought, bitwise, naught. Then inside these round dresses for now, I'm going to pass this circle. To show the example, we don't need to combine both the geometrical shape. Just we need only one geometrical shape. In our case, I'm going to use this circle and then C thought I'm from Matin. Then inside the round dresses, first, we need to provide the window nape. I'm going to copy the same variable leap as Window nip and then we need to provide the variable. And I'm going to set this file. After I set this file, also, I'm going to comment out this line, I don't need this line for now. I don't know how to show this one. So after set this file, if I run this code in my terminal, as you can see, this is Btis not operator. As I told you, it inverse the color. Not inverse the color. Basically, it turn zero into one and one into zero. That's why our something look black and our outside the circle area and the window area look white. This is it for this structurim. I hope you understand what is the usage of BTs operators and how we can use it. It is very essential to know these operators are very important when we work with us. Don't worry, we are going to learn about it in our project sections. So thanks for watching this video. Stay tuned for the next tudio. 16. Masking: Hello, guys. Good to see you back. Once again, I am back with another tutorial, related Python open CV. And in this tutorial, we are going to talk about masking. In our previous tutorial, we discussed about BTIs operator, and essentially using BTI operator, we perform Musk. In a simple sentence, muskin is used to focus on specified reason of an image while ignoring other positions. In open semi, muskin is used to focus on specified reason of an image while ignoring others, and this is achieved by creating a binary mask image where the other results of intersets are white and the rest are black. As you know, white represent with 255 and black represent with zero. And the mask can be applied to the original image to filter or manipulate the desired areas. Suppose if you have image of people and you want to focus on only their faces, in that case, then you can apply mask on their faces and hide unnecessary areas. Now let's see how actually it worked in open C. As you can see, here I read a file and display this image. And next, I'm going to import numbi moding because we are going to create a blank image using it. So I'm o type, import numPi as N P. Then I'm going to create a blank image. So here I'm going to declare a variable blank, an equal to Np dot zero. Then inside the round process to provide the shape to grab the shape, am type image, dot, shape. Then inside the square resis, we need to provide the dimension of the shape. Remember, the dimension of mask is very important. We need to take the same size of this image. Here I'm going to pass colon two and also we need to take the datatype for Tata type D type. Is single at Imo type I NDA. And if you want to display this blank image, you can just into type Cm dot, and for method. Then inside the round is hero type black and coma, and I'm going to pass the variable black. I will satisfy. Now I'm going to draw a circle on the blank image and I'm going to call the mask. Here, I'm going to create a variable called mask mask equal to mask, equal to C dot subcal. Inside the round recess, first, we need to pass the image where we are going to create and draw the mask. So pass black. Then we need to pass the center position, and to grab the center to define the center position, her type, IMG image, dot, shape, Is the square recess, one divided, divided by two. And next, image, dot, shape, is a square is zero divided, divided by two. And now also we need to pass the radius for radius for now, I'm going to take 100 pixel, and then we need to pass the color and I want to 55 ful white color as thickness equal to minus one to fill the circle with white color. And now we need to display this mask on image for that helm type. C V dot, I show then in such roundreses at first to provide a window name, which is mask. And then we need to provide the variable mask. In which image we want to create a mask, this one. And if you under scot, it's going to display tal three image. It's going to open total three window. Let me show you. First, it's going to open the cat image, then it's going to open the blank image, and this is our mask image. And now I'm going to use Bitwise operator to mask it. For that, I'm going to create another variable mask equal to CV dot bitwise and, Bitwise operator. Then inside the round resist, first, I am going to pass image as source, IMG Coma, image as second source, then we need to type mask mask equal to I'm going to pass this variable called mask. So using bitwise and operator, we apply this circle mask on this image. So if I display this image, so I'm going to type Ammtho copy this line and paste it here and replace mask with mask. They set this file and then code in my terminal. Let's see what it reted. As you can see, this is what it reted. This is our mask image. After I applied the mask, this is how it slow. At first, we took this image, then we put this image over this original image and create a mask. Basically bits and operator, we did the intersecting result of this area. If you want to move this mass, yes, you can, let me show you how you need to change the coordinates. So here, I'm going to uncoven unnecessary lines. I don't need it, so I comment out all of these and here, a divided by two, I want to move the mask, so I want to add fortified pixel. Here I add fortified in with these coordinates. After set this file and on this code, you can see, we move the mask area, we move the mask position. Instead of using circle, you can use a rectangle off. Just use it to replace circle with a rectangle. Just one thing you need to remember to applying mask. You need to take the actual size of image for black image. If you use any other dimension, suppose greater than the other bower than this, then it going to throw arrow. Using image shape, here we grab both the varies of image resolution. And if we change it, then it will to throw error. Let me show you the example. I want to duplicate slide and comment out previous one here instead of using the exact shape of this image, I'm going to take 300 by 300. If I save it this code turn in no, as you can see, it's 30. It's not going to work because we need to take actual shape of image to create the mask. I'm going to comment out this line and uncomment the previous line and set this file and run this code again. Now you can see it return the result. This is how we can apply mask. I hope now it's clear for you. Thanks for watching this video statute for the next tutorial. 17. Grey scale image histogram opencv: Hello, guys. Good to see you back. Once again, I'm back with a new tutorial related Python open CB and in this tutorial, we are going to learn histogram. Now the question is, what is histogram? A histogram is a chart that plots the distribution of numeric variable value as a series of bars. We are going to understand it when we start our practical. As you can see, here I open a Python file man dot Pi and I import a image, and I display this image using Imghofunction. Histogram basically allow you to visualize the distribution of pixel intensity in an image. Whether it's a colored image or whether it's a grayscale image, you can visualize the pixel intensity. You can understand pixel intensity distribution with the help of histogram. Which is like a grab or a plot. That will give you high level intuition of the pixel distribution in the image. We can compute a histogram for Griske image and compute histogram for RGB images. We are going to start off with computing histogram for Griskalimages. At first, I'm going to convert this image into grayscale image. For that, I'm going to declare a variable called gray gray equal to Cv dot CVT Color in the round dresses I passimge IMG Image variable, comma, and then I'm going to convert this ZR image into Grisker image, summary use CV, dot, color, underscore BGR, to gray. This one. Then I'm going to show this image in our window. I type CV. I'm going to type CV, dot, IMs method, IM so. Then you said the round dresses first I'm going to provide a display name which is gray. Then I pass coma and here I'm going to pass the variable, gray. Now, to actually compute the grascal histogram, what we need to do essentially, first, I'm going to declare a variable and our variable is gray, underscore hist, gray histogram, equal to CV taught, here I'm going to use a method, which is calc hist, calc hist, this one. Then inside the run recess, here we need to compute the historum for the image that we pass into. And here we are only interested to pass a grayscal image. I copy this variable gray and put it here. Apacoma, then we need to pass the color channel, and inside the square recess, I'm going to pass zero because we are using grayscale image, not colorful image. Now the next thing we have to do we need to provide a mask. So here, if you want to get histogram specific position of an image, then you can specify the mask position. But for now, I'm going to use the whole image, so I'm going to pass none. And next, we need to provide the t size and size is basically number of bits that we want to use for computing the histogram. When we talk about plot histogram, we need to understand the concept of B. But essentially for now, just set this two, five, six. Also we need to wrap this value as a list. The next thing, we need to specify the range of all possible values, all possible pixel values. For our case, inside the square se I'm going to pass from zero comma to five, six, and that's it. Now I'm going to import a module, matplot leap. Here, I'm going to import matplot leap as PLT. So type import Mc plot leave as PLT. Basically from this module, I want to import Pi plot leave. Here I'm going to pass McloteaptPi plot leave Pitht then I'm going to type PLT dot figure. I call this method, and next I'm going to type PLT dot title. Then in su roundre we need to set a level and our level is gray scale. It is a gray scale histogram. I'm going to type grey scale histogram. Then I'm going to set X label. So type PLT, dot, X level. Inside the rounders, I'm going to call these bins. Next, we need to set the Y levels. So type PLT, dot Y level. Here we need to pass number of pixel of pixel. Finally, we need to use the plot PLT, dot plot. And inside the round recess, I'm going to pass the very well called gray scale histograph gray hist. Also, we need to set the limit cross of the xs. I'm going to type PLT line. Inside the round recess, I'm going to pass the list from 022 56. Finally, we can display this image. Here I'm going to type PLT dot show Method. So I'm going to set this file. After I set this file, I'm going to run this code in my terminal. Up to run this code in my terminal, this is the image of histogram. It is the gray scale histogram image of this cat. Basically, this is the distribution of pixel in this image. As you can see, the number of beans across the Xxs represent the intervals of pixel intensities. So so as you can see, there is a peak at that position. This means that is close to 125, 125 bins. It's mean in this image there are close to 9,000 plus pixels that have intensity of 125. Basically, I want to say at that position, pixel intensity cross nearly 9,000. So this is essentially compute the gray scale histogram for the entire image. Now what we can do is we can essentially create a mask. Then we can compute the histogram only the particular mask and to create the mask, we need to import Numbi module. So here, I'm going to type import Numbi as NP. To create the mask, here, I'm going to declare a variable called blank, blank equal to Heuse a method called np dot zero. Np dot zero inside the roundreses here, we need to grab the image shape. Image dot shape and I want to grab to tell two values inside the square is colon two, the first two value. It's going to create a same size of blank image. Then we need to provide the D type value. D type equal to inside the single course UI NT eight. Then I'm going to create the mask using the blank image. So here, I'm going to type mask assigned width equal to CV dot circle. We are going to create a circle mask. Inside the round presses to draw this circle, first, we need to provide the image, which is black. Then we need to define the center position to define the center position inside the round presses. He to use image dot shape. Inside the square dresses I'm going to pass first value and it's divided with divided divided by two, we need to grab the vertically and horizontally center position. Egomotypeimage, dot, shape, Is that the square adress says, I'm going to pass zero divided, divided two. This calculation going to provide the center position of this image. Then we need to provide the radius of this circle, which is 100 pixel. Next, we need to provide the color. For color, I'm going to use 255 white color, single white colored and also I want to fill this circle. I pass minus one vu. This is the thickness value. If we pass minus one, it's going to fill the circle with white colored. And then I'm going to show this mask in N window. So I'm going to use Andromthod v dot IM and set the RunressF I'm going to declare a window name mask, mask, and I'm going to pass the variable mask. Now I'm going to create the gray scale histoum for this mask. I'm going to set this parameter mask parameter as mask. Instead of using nun Hemotype mask. But before I run this code, I think I did some mistake. It's not a mask, it's a circle. It is a circle. He create a circle. Here we create only this circle. We do not apply mask on this image. So to create the mask we need to use Bitwise operator. It is a very sllistic. So declared mask equal to CV dot and to use Bie endoperator bitwise and in said the rounds, first, I want to select the greyscalimage, this source as a source, gray as second source, also multi pass gray. And we need to have to pass mask, mask value equal to circle, circle. Here we apply this circle on the grayscale image as a mask. I'm going to set this file and run this code in my terminal. Up to run the scude in my terminal, here you can see the result. This is the mask area. And this is the histogram result of this mask image. Now you can notice there are a lot of spike on this histogram image because most of the areas are white. That's why pixel intensity is high on this image. As you can notice at 30 Bins position, we have 500 pixel intensity. This is how we can compute the grayscale image histogram. At the next tutorial, we will learn how we can use colorful image histogram. This is it for this tutorial. Thanks for watching this video, stay tuned for the next tutorial. 18. Colour image histogram: Hello, guys. Good to see you back. Once again, I'm back with another tutorial related histogram, and in this tutorial, we are going to learn how we can extract Cana image histogram graph. So as you can see, I already input our important libraries, CV two and Mt plot leave. Then I already read the image file using image it function. Also, I display this orgenic image using IM So Method. Next, we need to create a new figure window for the histogram plot. For that, I'm going to type PLT dot figure, and then we need to pass the round dress. After that, we need to set the title of histogram plot. For that, I'm going to type PLT dot title. Inside the round res, I'm going to pass the title nm and the title name is color histogram. Next, we need to take the X level name for that, I'm going to type PLT dot X level. Inside the round dresses, here I'm going to take beans. Level the Xxs as beams, where each been represent a color intensity value 02-55. At the same way, I'm going to take Y level. I duplicate this line and replace X with Y here I'm going to type number of pixel, number of pixels. The Y xs represent the color count of pixel for each intensity level in the histogram. Up to define title levels and figure. Next, we need to create a tuple. Here I'm going to tie a variable name and our variable name is color. Colors equal to, as I told you, it's a tupel so I'm going to use round briss side the round bris here I'm going to take total three tape value. Our first value is B, B for blue. Next, I'm going to take G. G for green, and our last one is R, R for red, R represent red. As you know, default opens V images in RGB mode. Here we define tape with color channel labels to indicate which color channel is being processed. After that, here, I'm going to take a follow flop to iterate all of the color channel. I'm going to type four, I comma call, call stands for color. And I stands for index in enumerate inside the rounds colors. Then inside this fall loop, I'm going to create the histogram, the histogram image. Hemo type a variable named HIST hist equal to CV I'm going to use a method called calc hit, calc hist. Then inside the rounds, we need to pass a list, and as a first list item, I'm going to pass the sequence, which is image IMG. This variable. Then we need to pass the channel, the color channel. Which color channel I want to target? One day what I want to target all of the color channel. I'm going to pass the index number. Instead the square adresses I'm going to pass I, then I'm going to pass the mask. For now, I'm not going to apply any mask, so I'm going to pass none here. Next, we need to define the hist size and for his size I'm going to pass inside the square sus 256. Next, we need to specify the intensity range to consider for the histogram. Here, inside the square ses, I'm going to pass 02-55. At first, we pass the source image. Then we specify the color channel, zero for blue, one for green, two for red, and I want to target all of it, here I pass I. I is the index number of color channel. I represent the index number and call represent the color channel limb. Now, the hist variable stored the calculate histoom values, which indicate the frequency of each pixel intensity in that color channel. Now we need to plot the histoom data for the current color channel. For that, herameertype, PLT plot. Then inside the round presses, at first, we need to pass a arc, arc like here as an arc, I'm going to pass Hite, this variable. Then we need to pass the color. Color equal to call variable. After that, we need to set X excess limit from 022 56 to match the intensity range. For that herramtygPLT, dot lime, limit. Inside the round press here to pass a list item from 022 56. Now outside the four loop, I'm going to display the histogram plot using plt doth method. I'm going to set this file. After set this file, if I run code in my terminal, as you can see, it open our histogram, color histogram. This is our original image. It's a colorful image, and as you can see in this histogram, we have total three color channel, red, blue and green. And if you notice each of the color channel, suppose for red at Essex win position, it touched the intensity peak for blue at 110 point it touched the peak. And at 155 XX position, you can see the green color channel Peak. You can see the distribution of each color channel in this grab. This process is useful for analyzing the colored distribution in an image and understand its overall color composition. Basically in this tutorial, we learn how we can extract colored image histogram, and in our previous tutorial, we learn how we can extract grayscale image histogram. Thanks for watching this video, stay tuned for our next tutorial. 19. Thresholding openCV: Hello, and welcome back to another video related Opens. And in this video, we are going to talking about thresholding. Now the question is, what is thresholding? Basically, thresholding is a binarization of an image. In general, we're going to take an image and convert it into a binary image where zero represents black color and 255 represent white color. Basically, we compare each pixel with the threshold value. If the pixel value is below then threshold value, then we set it zero means black, and if it greater than, then we set it 255 means white. In sense, we can essentially create a binary image just from a regular standalone image. And in this tutorial, we are basically talking about two different type of thresholding, simple thresholding and adapted thresholding. Let's start with simple thresholding. At first, I'm going to type a single leg command here. I'm going to type simple threshold. Simple threfolding. Before I apply simple threfolding, we need to convert this image, this colored image into grayscale image. For that, here, I'm going to declare a variable gray, assign with CV CT colored. This method. Then inside the round versus at first, we need to pass the source image which is IMG variable image, and then we need to use the colored code. Cvt color, underscore BGR two gray, this one. Then I'm going to display this image. So of type, CV do I am Somethor in the round brass, first, I'm going to provide a window name which is gray, and then we need to provide the image gray. Now to apply the simple thresholding, we need to apply a function. We need to use a function called Cv dot threshold, and this function return threshold. And also it return Trash. Equal to, then we need to use the method, vt threshold, threshold and inside the round dresses, first, we need to pass the source image which is gray, and next, we need to pass the threshold wave, a floating point. For now, I'm going to pass 150. Next, we need to pass the maximum value. If the pixel intensity value is greater than 150, then I want to binarize the image. If it cross the label, then I want to set it 255. I mean, fully white. Next, here we need to pass the thresholding type for thresholding type, I'm type C dot binary. This one. Another question is what it does. Basically it's going to compare each of the pixel value with this sash value. If it is greater than 150, then it's going to set it 255, meaningfully white. If it is lower than 150, then it's going to set it zero, which represent black color. Essentially it return two things TAS, which is the Hefold image or the binarized image. Also it return Tfold which is essentially same value what we passed here. In our case, 150. Now, let's display this image, so type, Su type dot and Mthods the Rundverses at first, we need to provide the window name type, simple threshold, and then we need to pass the images this one now let's run the code in a terminal and C to work propeller naught. As you can see, this is our original image. This is the grayscale image. This is the griscal version of this original image and this is the threshold image. This is essentially we get, if we say threshold value 150. Now, let's change the threshold value. I did this terminal and I'm going to change, I'm going to make it 100 threshold value. Also, I don't want to show the original image. I comment out this line and also I don't want to show the gray image. Also, I comment out this line. I'm going to sub this file. After set this file, if I run this code in my terminal, now you can see this is what it's written. If we say thrash value 100. Now, the pixel value above the 100 become 255 and the pixel value below the 100 become zero. Instead of passing higher value, if I pass lower value, let me show you. If I pass 220. Then set this file and run this code in my terminal, now you can see the different result. Now you can see only two eyeballs, nothing else. This is how you can get different result if you pay with threshold. Now, let's inverse this threshold of two. For now, I'm going to make it for now, I'm going to make it 120 and set this file, then I duplicate this section. Then I'm going to change the even pra underscore inverse, I NV. Also, we need to change the method. Cv dot binary inverse, N. Also, I need to call this variable Trash underscope NV and simple threshold NV. This is the window name. Basically, it's going to inverse the image. Image pixels, let me show you. After set this file, you can see the result. This is the original simple threshold, and this is the threshold image with inverse mod. As you can see it inverse the pixel values. Now, white portion becomes black and black portion become white. This is the inverse, the version of this image. Now, instead of setting pixel intensity which is greater than 120 into 255, it's going to set it zero. Because here we inverse the image. We use thrash binary inverse. Basically, you want to say which value is larger than 120, it going to set it zero and which value is lower than 120, it going to set it to 55. This is simple threshold. Now let's move to the adaptive threshold. Here I'm going to type a single encoman a stack, and I'm going to type adaptive thresholding. In simple threshold process, we need to provide the threshold volume manually. But in adoptive threshholding, we left the computer to find the optimal threshold value by itself, and using that value it binarize the image. And to contact the adopting threshold, here, I'm going to declare a variable called A underscoe A equal to test for adoptive thrash, equal to Cv dot here, we need to use a method called adaptive threshold. This one. Then then inside the round ss, at first, we need to pass the source image. Here I'm going to pass gray, the gray image. Then also, we need to say the maximum value. We don't need to set the thrash value by manually. We need to say the maximum value and I'm going to say 255. Next, we need to use the method adaptive method, which method will apply for the adaptive trash. I'm going to type v dot adaptive trash main C, this one. Also, we need to call threshold Ti, which is Cv dot binary, this one, then we need to pass block size for block size, I'm going to pass 11 at last, we need to pass CV and I'm going to set Cvu three. Now, what does that mean? The function called adoptive threshold is a thresholding method, which is very useful for images with varying lighting condition across the different area. If I talk about this parameter, C adoptive mean undersco C, this parameter define the method used to calculate the threshold for each pixel. At the next parameter we define the thresholding time. Here you can see her we pass everywhere. It is for block size. I represent the size of neighborhood area. Here 11 represent 11 by 11 block up pixel will be used. At last we pass three. I represent weightage. It helps fine tune the threshold. Method these methods helps in handling images where lighting may very significantly cross different areas as it adapts to local variations. Now, go ahead and display this image. Here I'm type Cb dot I'm from method. And inside the rundresses I'm going to pass same variable name as window name as adopted and comma I'm going to pass the variable. Also I want to comment out these two line, this one, and this If Math I'm going to display the gray image. I'm going to set this fine. Now, let's run the code in our terminal and see what it return. This is what adoptive Tash return. This image. Here we define the kernel size 11 w 11. And OpenCV computes a main over those neighborhoods. I want to say neighborhood pixels, and it's find the optimal threshold do for the specific part. I know it technically sounds crazy and it's difficult to understand what is going on behind the scene. So this is how we can apply adaptive threshold on image. Threshold going to play a key role in our projects in our upcoming projects. That's why we need to understand how threshold works. So thanks for watching this video, Syuned for our next tutorial. 20. Displaying the coordinates of the points clicked on the image using Python OpenCV: Hello, guys. Good to see you back. Once again, I'm back with another exercise related Python open C. Today in this exercise, we are going to display the coordinate point when I click on the image. Suppose if I press right click on this image at that position, as you can see, it print the coordinates X XS and YXS coordinates. And the same way, if I praise right click on it, you can see also it print the coordinates. Let's see how we can achieve this program using PyonopenC. So finally, we are in my Visal studio code editor. OpenCV help us to control and manage different type of mouse events and give us the flexibility to operate there. There are many type of mouse event. This event can be displayed by the running this code. As you can see, Hey use a one liner loop for I DIR function CV two. I eve P in. It's going to print all the events from the OpenCV, from the open CVrator. Let's run the code in my terminal and see how many evate we have. Up on this code, as you can see, we have multiple keyboard and mouse event. Flack old key, control key, flag lip button, mouse button, R button, shift key, but we need to display the coordinates when I click on this image. For that, we need lift button down and write button down, R button down, this to event. Let's start the code. So first, I'm going to kill this terminal. Then I'm going to comment on this line. Next, here, I'm going to Read and image using image root function, IM read. At first, I'm going to declare a variable called image IMG, image equal to CV two ImtFunction. Im read. This one. Then you said the round presses. At first, we need to provide the file path, the image source. Is the single course, I'm going to pass cat two dot JPG. With that, we need to pass the color channel and our color channel mode is BGR. I want to pass one. If I pass zero, then equal to return grayscale image. Next, we need to display this image using IMS function. He type, semi two dot Im instead the run press, first, we need to provide a window name which is image. Then we need to provide the variable, the source image IMG. And now we need to set a mouse called that function. We going to pass the image window and use a defined function as parameter. Here modo tie CV two dot set mouse, call that. Then inside the rounds, we need to pass the image name, the image window name, not the variable and the image window name is image. Is the single code, Imo type image. Next, we need to pass a user defined function as parameter, and our function name is click event. So type, click, underscore key event. This function going to handle all the mouse key events. Also it's going to return the coordinates. Next, I'm going to call the wet so type CV two dot, weit key. Wet key is a d zero. I'm not going to explain you why we need to use Quit key because we already learn about it in multiple videos. Also, I'm going to destroy all the Windows. So type CV two dot, destroy all Windows. Next, before I create this function, I want to move all the code inside the if named equal to equal to mean function, which is our driver function. I'm going to type I ndscoe underscoenme, undsco underscore, equal to equal to inside the double codes undersco underscoe M, undersco underscoe. Then inside this driver function, I'm going to move on the code. We use if naan equal to equal to main function because we are going to call this function from the existing page. That's why we need to create it. Now I'm going to create this function, click event. First, I'm going to copy the function name, click event, and there so here T D our function name, click event. Then inside the round presses, this function going to take multiple argument. Our first argument is even. Our next argument is Xs. Then I'm going to pass Y. Also, it's going to take flag. And last one is pardon. Then inside this function, for now, I'm going to type pass. For now, I think I need to print something in a terminal. So I'm going to type print statement, print. Inside the process, I'm going to type, it's work. Before I run this code, let me explain what we did. As you can see here we use a function from OpenCV, cv two dot, set mouse callback. It's a mouse called that function. Wherever I click this window image window, then it going to call this function, click event. And in our click event function, we just print its work in our terminal. If I set this file and run this code in my terminal, you can see the image. And whenever I click this image, as you can see it print, it works. Basically, whenever I hover my mouse on it, also it it work because in our function, we do not specify which particular event we are going to use. That's why it work on any events when I hover my carton on it. Also, if I press any key, I think it's still going to work. No, it's not going to work if I press key because here we use only mouse call that function. Now let's talk about the parameter of this function. As you can see, we have total f parameter, X, Y, flax and params. Using event parameter, we can define the mouse event. It may be lip click, otherwise, right click. X and Y, the coordinates open mouse event. Then come flax parameter. It is not used, but but it required by the callback signature. I want to say any relevant flags passed by Open C. And then come patterns. Pattern is also additional parameter. If it's necessary, then we are going to use this parameter. So first, I'm going to use IP condition in this function. So type, we are going to checking for the mouse click event. I even equal equal to CV two dot event here I'm going to use a event, and I'm going to use lit button down event, some type event. Underscoe L, button down, this one. Then inside this IP condition, first, I'm going to display the coordinates, and I'm going to display the coordinates in our terminal. Subtype prin inside the rounds, X I'm going to provide a space, so I pass the whole codes, I'm going to pass Y coordinates. Then I'm going to duplicate this section. I want to study IP condition. Because here we need to secret two IP condition. One is for it button down and another one is for write button down. I'm going to set this file. Now I'm going to run this code. Run this code in my terminal. If I click on anywhere on this image, let's see what it written in my terminal. As you can see, it print the coordinates. First, it print the xx position, then it print the y position. If I praise red kick at that position, also print the coordinates. It is for lip kick and red iOs because we use the IP condition. So we're almost done. We need to print these coordinates on the image. So at first, I'm going to use font. For that, we need to use fonts, sort of Ti, font, EPI two. Here I'm going to use font from OpenCV, CV to do, and here I'm going to use font, underscore, He semis, this one, this one. Now I'm going to use a inviil function from OpenCV, which you going to put the text on image. For that, I'm going to type CV to that Put text. Inside the Put text function, at first, I'm going to pass the source image, which is IMG. C then I'm going to pass a string. So I'm going to call string function TR. Then inside the round verses, first, I'm going to print X, X. Then I'm going to concatenate with Cgation sine. I'm going to pass inside the single codes C. Then again, I'm going to use congatenation sine again, I'm going to print a string for Y VD to print the Y VD. STR, Is the rounds is Y after this stream, we need to define the position where we need to place the text. And for that, inside the rounds, I'm going to pass excesses and Y xs. Somnotype X, Y. Then we need to pass the font phase. Which font I'm going to use. I'm going to use this variable form because in this variable we define the font. Last, we need to define the font scale, which is one, then we need to define the color font colored, and I'm going to use BGR. I said the runbress and I'm going to use blue colored. 255, the first value, second value is zero, also third value is zero. Then at last, we need to define the thickness of this font, which is two. That's it. Then we need to display this image using IM show function. So I'm going to copy this line, and I'm going to paste it after that. Also, I'm going to copy this section and paste it for this condition for the right key. I'm going to set this file. After set this soil, let's run the coat and see is it to work properly or not. So up this cod inout terminal, if I click anywhere, so fast I'm going to praise let click. UterPrais let click, as you can see, it prints the coordinates on the image. X is X excess is 130 and Ys is 170. Also it prints the same value. In our term. Now I'm going to praise Rattle. If I press R cli as you can see, this time also print XXs and XS coordinate. This is how we can print the XXs and xs coordinate using OPC. I hope now it's clear for you how we can create. Thanks for watching this video, Schedule for the next project. 21. White and black dot detection using OpenCV: Hello, and welcome to another exercise real tt Open City. In this exercise, we are going to learn how we can count that dot and white dot using Python open City. So let's see the demonstration. As you can see in my screen, side by side, HeaOpen total two different image. First image come with white background. And in this image, we have multiple black dots. And the second image come with black background. And also, in this image, we have white dots. But if you notice, you can see the dot size is different. Some doors are three squared pixel and some doors are nearly 20 squared pixel. So we need to write a program using Open CD, which is going to count all the black dots, A which is going to count all the white dots. So let's jup into the visor studio code and see how we can create it. Hello, and welcome to another Python Open Cit exercise. In this exercise, we are going to perform White and Black dot detection using Open CP. Image processing using Python is one of the hottest topic in today's world, but image processing is a bit complex and beginner get bored in their first approach. We have very basic image processing Python program to count black dots in white surface and white dots in black surface using OpeCVFunction. In this exercise, we are going to use function like image read threshold, fine contours and contour area. As you can see in my current working directory, we have a Python fine mean dot pi. Also, we have a image black dot one dot GPH. This is the image we are going to use to find black dots in white surface. As you can see in this image, we have multiple black.in white surface. We need to count the black dots using OpenCV. So let's get to the Visual Studio code editor. So first, we need to import the OpenCV lining. So here we type, import, import C two. Then we need to define the image five bath, which is black dot one dot GBG. I'm going to declare a variable path equal to inside the Dual codes, black, dot one dot JBG. Now, we need to convert this image into grayscale image. For that, I'm going to use image at function. First, I'm going to declare a variable gray, assigned with, I'm going to call a function C two dot IMRdimageRad. Then inside the undresssF only to provide the image file path. This one then we need to provide a tu flat value, which is Z. This can convert any image into black and white image. Now, we need to convert this griscal image into a binary image. For that, we are going to use threshool. Type a single and command threshold and hen type TH and another variable thrust. S equal to CV two dot ahold. Then inside this function, first, you need to pass the grayscale image, this one. Next, we need to pass to. First, we need to pass the thrash B. Here I'm going to pass 100. Then we need to pass maximum level win, which is 255. Then we need to pass the type, which is binary. I'm going to type two dot thrash Trash binary IN V. Then I'm going to use Odd operator and here to type CV two dots underscore forts, this one. Notice this file. Now, let me explain the threshold function. The function CV two threshold work as if pixel value is greater than a threshold value, it is assigned one, maybe white, else, it is assigned another value, maybe black. Basically, it's going to convert this image into black and white image. I want to say pure black and white image. You can call it binary from of an image. The first argument is image source, which should be a grayscale image. Second argument is the threshold value, which is used to classify the pixel values. If the image pixel value is less than 100, then it's going to set zero. If the image pixel is greater than 100, then it's going to say 255. As you know, zero stands for black color and 255 stands for pure white color. C two, under scoe binary uncoe IND is a threshold time. It's mean the output binary image will have the inverted color and OTsU is used to automatically determine the optimal ThessalVdo. I know it quite sounds technical. Don't worry about it. Just do what I'm doing. Next, we need to find the condues. Now the question is, what are the condues? Condes can be explained simply as a curve joining on the conduce point along with boundary. Keep the color and intensity is same. The condues are a useful tool for shape analysis and object detection, recognization, and menu. Contours give better accuracy for using binary images, and the find conduit function take total three argument. First one is source image. Second one is contuRtrival, and third one is contour approximation and fun condus return a Python list. Let's use the function. First, I'm going to declare a variable, CNT equal to condues equal to CV, fine conduFine. This one, fine contues. Then inside the roundressF we need to pass the source image, which is thrash, this one. I copy this variable and paste it here. Then we need to pass the retrib moon. For that, I'm going to use C two dot ETR ist. Retv list. And the third value is approximation method. For that, her pass CV two dot chain approximple. This one. The chain approx sample method comprise horizontal, vertical and diagonal segment and leaves only the end points. This results in fewer points being stored in memory, which is useful for simplifying conduit representation. Then next, inside the square versus m to pass minus two. This is used to handle differences in the written values of Cout fine conduce method. It is used to control differences in different version of open C. In some version, it return total three value, contues hierarchies and image, and in some version, it return total two value, only contut and hierarchies, and especially we need contours and hierarchies. We don't need images. That's why hero use slicing technique minus two because we need to select only this two value, conduce and hierarchies. So this is going to find the on the contest and store this contour in the CNTs value. Now, we need to filter this confuse. We need to filter those condue based on their area. We need to keep those confuse whose area 3-20. I want to say three and 20 square pixel. After filter this conduce, we need to store this contule in a variable. So let's get the filter. So here we need to filter the contube by their area. So first, I'm going to define a one. This is our first feeder point, which is three square pixel. Let's I duplicate this line, and this is our second feeler point, two, which is 20 square pixel. Now I'm going to create a empty list, which is X CNTs X contubes equal to, it is an empty list. And now we are going to run the loop through all the contues. For that, I'm going to use 44 CNT, contuN conduce, for conduce in contours variable is with all the contes then at the fall He use p condition. E is one, I S one means condition one, three square pixel, less than CV two. He to use contour area method contour area. Then inside the rounds, we need to pass the cones CNT. I mean if each of the contour match this condition, N contours and with that, it is less than S two, me condition two, with inside this tip condition, I want to append all the contours which is match this condition in this list. Basically, I want to append all the tues in this list, which is match this condition. Hemo type N X conduce append function. Inside the run press, I'm going to pass the particular conduits, CNP. That's it. Now, let's explain this form. First, I run a follow to all of these contours which we store in this list, tu list. Then each of the contour makes this condition and the condition is in the contour is between three square pixel to 20 square pixel, they append this contour at at least X contues. And her create the Dist conduce. Now we need to print the length of the conduce to print how many.in our image. For that, outside the fur lo amtype print statement, print inside the round brasses, inside the double codes, Hemotype total number of that dot. And here, I'm going to use the Estring method. E string, then inside the curses, I want to print the length of this list. So I'm going to use the length function AN and inside the round recess, I'm going to pass exent the list. I'm going to set this file. And before I run this code, let's check the code if I did some mistake or not. I think it's fine. Oops, there's a mistake. Yes, this one. We need to move this selection outside the round braces. So I cut this section and put it outside the round braces. And I'm going to set this file. And after I set this file, let's run the code and see easy to work properly or not. So to run this code, as you can see, the total number of black dot is 23. So it worked perfectly. So we have totaled 23 black.in this white surface. So I hope now it's clear for you how we can count black dots in white surface using Python Open C. Using this program, you can count black dots. Also you can count the white dots. Let me show you. So as you can see, we have another image in my current working directory. This image come with black background, and in this black background, we have white dots, but we have hundreds of white dots, and white dot size is different. Some white dot is nearly three square pixel and some white dot is nearly 20 square pixel, we need to count all of them. Let's duct the users studio pod. So as you can see in my current working directory, we have this image, white dot dot pinch. First, I'm going to copy the file path, otherwise file in. Then I'm going to replace the file path, so I'm going to duplicate this line and comment on previous line and replace that dot with white dot png. Next, I'm going to change the total number of white dot. Es, also, I'm going to run this same program, but just only to change one thing. This time, I don't need this inversion. I just need thresh bind. So I'm going to duplicate this line and comment out previous one and remove INV. We just need thresh bind, and I'm going to set this file. Up set this file. If I run this code in my terminal, let's see what it written. Up run this code, as you can see total number of white dots, 583. So this is how we can count the white dot also. Just we need to change the thrash bind. If you want to count the black dot, then you need to invert the thrash bind, and if you don't, then you don't need to invert it. So this is it for this tutorial. Thanks for watching this video Stay tuned for the next tutorial. 22. Draw rectangular shape and extract objects: Hello, guys. Good to see you back. Once again, I'm back with another Python open CP project, and in this project, we are going to draw a rectangle shape and extract object using Python open Civing. Let me show you the demonstration. As you can see, here I open window of cat image. Now I'm going to select some area of this image. I praise lip click and release at that position. As you can see, here it create a rectangle shape. And if your selection is not well, also you can restore the window. Also you can reset the window, you need to press R. Again, you can make this selection. And if you think your selection is good, then you need to press C keyword in your keyboard and it's going to crop this section and display it in a new window. As you can see, Harry paint crop image, and it display the image at that window. If you want to quint this window, just to press Q to quote the program. And to run the program, we are going to use common line argument. This is what we are going to create in this project. So let's start the practical. Here we are going to write a Python code which allowed to user to select and crop portion of an image by drawing a rectangle using mouse weds. At first, we need to import modules. First, I'm going to import OpenCV, sound type CV two. Then I'm going to import another module which is arc perse. I'm going to type import Rc parts. So what is the usage of Rc peruse module? Rc peruse module is used to handle command line arguments. Don't worry, we are going to learn about it later. Now, let's initialize the list of reference point. Here I'm going to create a empty list, and for that, I'm going to take this variable name RF underscore reference point. Reference point equal to it is an empty list. Then I'm going to declare another variable crop, by default, Cp equal to Fs. This condition is forbs. Reference point variable is empty list. Which going to store the starting point and the ending point of the rectangle. And crop is the Boolean value, which is used to track if cropping is in progress. Next, we are going to use to mouse event function, which is going to track the mouse clicks. For that, we are going to use basically two mouse event, which is lap bottom down and lap bottom up and we are going to use this to mouse event in a function. Hem drew declares function F and our function name is shape selection. Shape underscope selection. Shape selection inside the round dresses. First, it's going to take event as parameter. Here we need to pass the event. Event, then we need to pass Xxs and YXs. Next, I'm going to pass the flags. At last, I'm going to pass pattern. Inside this function, we need to use these two variable. Here I'm going to declare this two variable as global variable, some type global reference point and the next variable is crop. Now, if the late mouse button was clicked, record the starting X coordinates and Y coordinates and indicate that cropping is being performed. I want to say if I click the lap button on the window, then it's going to grab the starting rectangle point. So we need to use IP condition event equal to equal to here I'm going to use mouse lip button event. I'm going to type C two dot event underscore, and I'm going to use lip button down event button down, this one. Then inside this IP condition, I'm going to say the reference point wave reference point equal to here I'm going to say it inside the squareds, inside the rounds, I'm going to set the first value. The first value is X, Y. Here we use if event equal to equal to let button down, then it's going to grab the Xxs and YX and coordinate and save it in this empty list. Now also we need to grab the end point for that hem type ellip condition. Else if event equal to equal go. This time, I'm going to use CV two dot, even, underscore, and this time I'm going to use lipped button, p button. I mean when we release the lapped button, then I want to grab the endpoint of the rectangle to grab the coordinate, we need to use apenFunction. We need to append another xs point and YXs point to this empty list. For that, I'm going to type reference point dot apen. When you said the roundss, I'm going to pass the coordinates, coordinates of XX YX. Now we can draw a rectangle because now we have the starting point and the N coordinates. Is the function and outside the condition, I'm going to draw the rectangle. So type C two dot rectangle. Then inside the roundses in our first parameter, we need to pass image. In the second parameter, we need to pass the first coordinate. Xxs coordinate and Y excess coordinate together. To select the first coordinate, I'm going to tie reference point reference point. Inside the square resis, I'm going to sleep the first view. So I'm going to pass zero. And then at the cooma again, I'm going to type reference point. Inside the square resis, I'm going to pass one. It's going to select the second view. Where you already store, the Ss and Xs coordinate. After that, we need to select color, the color of rectangle. Inside the round dresses, we need to provide color as GR format. For blue, I'm going to pass zero. For green, I'm going to pass 255. And for red, again, I'm going to pass zero. Basically, I want green color. Then we need to pass the thickness of the rectangle border. I'm going to pass two and then I'm going to show the rectangle in this image. For that, I'm going to use IMS method, sometimee CV two dot IMsO then inside the round presses, I will pass inside the double code, need to pass the window name which is image. Then we need to pass the image variable. Image. We write a mouse event function which can grab starting point of rectangle and endpoint of rectangle and also it can draw a rectangle on an image. And now we need to construct the argument parser and parse the argument. Is the function. At first, I'm going to declare a variable called AP. AP equal to heard I'm going to call this module arc parse type R parse. From here, I'm going to run a function called argument parser. This one. At the next line, I'm going to parse the argument. For that, I'm going to type AP dot here I'm going to use another function call add argument. Add underscoe argument. Add argument inside the rounders, here at first, we need to pass a screen. Wh going to work as name, otherwise, flax. Inside the double course, I'm going to pass I. Then outside the quotation, we need to pass the same argument. Inside the double course, I'm going to pass image. Com this argument is very important, so I'm going to make it required. Required, equal to true. And to understand the line, also, I'm going to type and to understand the line for what purpose I type this line. I'm going to add a hell. Hell equal to inside the double course, I'm going to type. This is the path to image, so type path to image. Don't worry, I'm going to explain it step by step. Next, we need to parse the argument. For that, I'm going to type another variable args. Args equal to, I'm going to call a function called words. Words inside the round ress I'm going to pass apt parse arc. I'm going to call a function parse arc. Now let me explain you what is that? Why I use these functions. Basically, it is construct an argument parser using the arc peruse library. At first, here we create a argument parser object. The initialization and argument parser object, which will handle command line argument pass to the script. Next we add an image argument. I and image, these are the short and long option for the argument. The user can specify either when running the script, and then we pass required true. It's meant the argument is mandatory. If the user does not provide the image path, the script will through an error. Next we pass L basically this provide description of an argument, and at the next line, we parse the argument. First, we talk about apt pass Rg. This method parse the common line argument and return an object. Then we use a function called verse. The verse function convert parse argument into a dictionary. You can easily access the value by their argument name. Basically, when we run this script from our terminal, then we can directly provide the path of an image. And now we need to load the image and clone it. Also, we need to set up the mouse Globe function. Let me show you how to load the image, I'm going to create a variable name image. Image equal to. Here I'm going to read the image using image read function. So type v two dot image read, I am read. Then inside the round presses, here we need to provide the image file path to read the image. But as I told you, we are going to pass the file path when we run this script as a argument, and we already did it using arc purse method. So from the arc purse method, I'm going to pass RCs this variable and inside the square recess, we need to pass the image. Then I'm going to clone this image, so type another variable clone equal to image dot copy function. Next, I'm going to assign a window name for this image. Sono type CV to name window, name window, Window. Inside the roundres I'm going to pass the same name image. And now I'm going to set the mouse callback function. C V two dot set mouse call back. Inside the run process, first, we need to pass the window in which window, I want to set the mouse callback which is image. Then we need to call an event in which event we want to call and I want to call this event shape selection. Copy the event name and paste it here. Now let's check it and is it to work properly or not. Here, I'm going to type C two, dot, quite key. Inside the round process, I'm going to pass Z and also I'm going to destroy all the windows. I'm going to type C two dot, destroy all the window. Puts code in my terminal. As you can see, I didn't mistake the spelling of. It's E, not A. I'm going to set this file. If I run this code in my terminal, as you can see, it's going to return an error because Harry do not pass the argument. But if I pass the argument, I want to say the command line argument. If I pass the command line argument, so I'm going to type the script, some to run the script, some type Python. Space, the file name M dot Pi. Then I'm going to type ID next I'm going to pass the FlePAT. As you can see, our file path is cat two dot JPG. Here I'm going to type cat two dot JPG. Then I'm going to hit Enter. After I hit Entered, I think there is an error. Still there is an error. But it working. As you can see, I can draw the rectangles. As you can see, it's written in index error, least index out of fringe, line number 18. First, to resolve the problem. I'm going to delete this terminal and then jump into the line number 18. Yes, I think I identify the problem. The main problem is we need to move this two line inside the lip condition, not outside the lip condition. So I'm going to move it inside the lip condition and the set this file again. If I diggle another code in my terminal, it's going to throw an error. But if I tie Python, if I pass the command line argument, main dot pi I, then I pass the image name gat two dot JPG, I hit Enter. As you can see this time it work perfectly and it don't throw any error, and as you can see, we can still draw rectangle using our cursor. Our mouse events work perfectly, and it do not throw any error in our terminal. As you can see, we can draw multiple rectangles. Then click the mouse, I assume as a starting point and when I release the mouse, mainly the left click, then it's going to assume the release point as a ending point of the rectangle. We done our 80% of work. Now we need to crop the selection area of this image. Let's do it. I'm going to read the terminal also, I'm going to comment out both the lines, otherwise, also, we can remove it. Now I'm going to run a loop which is going to display the image infinitely until we praise Q. Here I'm going to type file. Two. Then inside the Value, at first, I'm going to display the image and wait for keypress. For that, I'm type V two, do I am so method, I am so. Then inside the round dresses, I'm going to pass the window name image. Also, I'm going to pass the source image which is stored in a image variable image. Next, I'm going to assign a variable called key. Key equal to CV two dot wet key. Inside the round presses, I'm going to pass one. Now the quoti is YI pass one. It's going to wait key praise for one millisecond. If a key is praised, it captured the key code, with that, I'm going to pass a bite, a bitwise code, zero X if H. This bitwise operation ensure that only the last eight bits are considered. Best ensure cross platform compatibility for the capturing key coodes. Now after I create the rectangle, if I press rkyword, then I want to reset the window. For that, we need to see a condition. We need to set IP condition. If key as I told you, it's going to capture the key code. Key equal equal to her to run a function called ORD, or inside the rundres inside the Dugal codes, R key, then inside the IP condition, I want to clone the image. I'm going to type image equal to clone. Dot copy. Next, if I pass C, then I want to break the loop. Hemotype condition, LC K equal to equal to ORD, is at the roundres is at the quotation, C, then break the loop. Break. Now the question is, what is RD function? ORD function convert the R character, otherwise C character into a AI V I allow comparison with the key code, which you capture from the Wek function. Now outside the Y loop, we need to crop the image. For that, again, I'm going to use I condition. So root I here I want to call the len function AN inside the round verses, I'm going to use the called reference point. I copy this reference point and paste it inside the len function. If length of this reference point equal to equal to two, then insert the condition. Crop the image. So type variable, crop underscope image. Crop image equal two, and here I'm going to use the clone image. I'm going to copy this variable clone. Then inside the square says, we need to define the crop position. But let me explain what is reference point equal to equal to two. Why I use this condition. Up the user phis selection of the area and create the rectangle and ater pressing C to bake the loop. The code going to check the reference point hold two value or not, and the two value represent the starting point and the ending point of the rectangle. Then from the clone image, we need to define the cropping area to crop the image, we are going to use slicing technique. Here Imo type, reference point R underscoe point is at the squares, I'm going to pass zero. Then again, I'm going to use square resis here I'm going to pass one. Then I'm going to use colon. Don't worry, I'm going to explain the points. Again, I'm going to type reference point. Is that the square addresses this time I'm going to pass one and one. Co the next value, I'm going to pass for weed. Reference point, I at the square resis, I'm going to pass zero. Also, I'm going to pass zero. Colon, reference point inside the square s is Helm to pass one and then inside the squares is I'm going to pass zero. Now let me extend the coordinate this slicing technique. This selection is for height selection. It's going to select the area from the YXs direction. It's work row is. It's going to select from the YXS point and ending YXS point and the next selection is for column wise. It's going to select the width. First, it going to select the starting xx point and then you're going to select the ending XS point. Then I'm going to display this crop image in a new window using IMshoFunction. Here I'm going to type cv two dot Imo inside the run presses first fully to pass the window name which is Crop Image. Crop underscore image, comma, then I'm going to pass the variable crop image, the exact area I want to show, and also I'm going to run ct C two wet key Heunpass zero. Updo that inside the IP condition, I'm going to destroy all the window. For that, I'm going to type CV to dot, destroy all Windows. That's it. I'm going to set this file. Now, let's set the file and if I run this code directly in my terminal, as you know, it's going to throw an arrow. I'm going to run this code from the command line. First I'm going to type Python space, fine name main dot pi dot pi space I space, and then we need to provide the file path, which is cat two dot JPG. If I hit Enter, as you can see, it show the image. Now, I'm going to select some area from this image. I'm going to select this is our starting point and this is our endpoint. I select this area. I think I don't like this selection. I want to remove this selection. For that, as you know, we already use a Kort, which is R. If I press RKOd it's going to remove the selection. If I select the image multiple time, also if I click R, then it's going to remove this selection. But after select the area, if I praise, C, then as you can see, it crop this section and display it in a new window. This is how it's worked. I hope now it's clear for you, how we can create this one. Thanks for watching this video, stay tuned for the next tutorial. 23. Text Detection and Extraction using OpenCV and OCR: Hello, guys. Good to see you back. Once again, I'm back with another project related Python open CV. And in this project we are going to perform takes detection and extraction using Python open CV. As we're going to take help of ORC. I want to say Tsat ORC. As you can see in our current working directory, there is an image named simple dot JPG. If I open this image, as you can see in this image, we have text. We have a sentence. This is a simple text. Also, you have another sentence. Text is at different reason. And now we need to extract the text from this image and we need to save it in a text file to extract the text, we are going to take help of Tsat Oc. Now let's see what is Tsat Os and how we can download it in our local system. Before we start our next project, we need to download Tsat OC and we need to install it in our local machine. Open Google and search for Desert OC. Then it will provide a link and you need to click on the first link TesarC OpensurceOC engine. But before I open this link, let's talk about Tsat OC. Test is an optical character recognization zin for various operating system. It is a free software released under the Apache license. Basically, it is a optical character recognizer. Using it, we can extract text from a image. Before we start the project, we need to install it in our local machine. I'm going to click on the first link. It will redirt to a Github web page. TaserecOC is a common line tool. We can run it through our common palette and using it, we can extract text from an image. Now you need to jump into the installing section, scroll down a little bit, and here you can see this section, Installing Tresorat. Then you need to install TrasatYPbuild binary package. Just click on this link. It will redirect to this page. From this page, you can found the different version of this traseract. Also you can found different operating systems. I am Windows user, so I'm looking for Windows. If you are Mcuser then use this link. So I'm a Windows user. That's why I jump into this section. Then you need to click this link, Taser at UB Minh link. If you click this link it will provide 32 bit. As it provides 64 beat. I click this link and it redirect to this page. I am using 64 bit operating system, so I'm going to click this link Tsat ORC, W 64. I want to click this link and it start my download. Now after finish the download, you need to open the folder and you need to install it as like other software. Then you can directly open it. Otherwise, you can open it run as administrator. Then you need to select the language, then click Next and you need to follow this simple process. Next Mix next. I agree. I want to install it for everyone who using the computer, so I click next and then just press mix. Also from this section, you can select the destination folder and I'm going to leave it as it is. Then I click Install and start the installation process, and our installation is complete. I click Next again and click on finish. Then we need to found the installation folder. I click, so I'm going to click Crve and then I'm going jump into the program files. You can see we found the folder called TsarcORC. If you open this folder, here you can find all the necessary file for Trasact. Then back to the program file and at first, I'm going to change the name of this folder. I just want to name it Tsat. I'm going to remove dash and ORC. I try to name it Tsat. And then I said this name. Now we need to open this folder ag and I'm going to copy the file location. I click on this file path and I just copy this file path. Then you need to add this file location to your path. For that, search for PAT and you need to open these environment variables. Then just click on environment variables, and you need to select the PT section. Then just it and add a new Path. Now you need to paste the Tsat file path. Set this one then you need to open your Windows power shell, and you need to tie Tsat. TE WSE RACT. Tasert hyphen hyphen hell. If I press Enter, let's see what it returns. As you can see, it's working. From this command line, you can directly use Desarat and extract text from a image. But we're going to use Python programming language and create a sophisticated program. In this section, we learn how we can install Disac and how we can set it as environment variable. We are going to use Dsarec with Python. Also we need to install a module, so tie P, install space Pi desert, PI TEs E RA CT. It's not sat, it's Tsat. I'm going to remove R and replace with T. Then just install this module in your local system. It is necessary for our project. I hit Enter. After hit Enter, as you can see, it download this module and install it in our local system. So after install TsaicOs and set it as a environment variable. With that, we already install a module, which is Pitste. Using the help of Pitsact module, we can run Tsaric functions in our Python file. So at first, let's import the modules. At first, we need to import OpenCV module, Stipe, import CV two. Then I'm going to import Pisact module, Styimport Pi Tsat. Then we need to mention the tasaic location in our local system and we install Tsaric in our PoormFle. So to pass the name of this Poa, which is Tsat. Here type command from Pisaic mode. 40, PisacPiTsac dot Again, I'm going to type Pi tasact dot. This time I'm going to run a command Tsact CMD. Tasert underscores CMD here, we need to mention the location. Inside the single code, I'm going to type the folder name, which is only Taserc not tasaic over C. In a small case, I'm going to type taser. As you know, TsariT is a common line program. That's why we need to use Tsat command. Then we need to read the image. In which image, I want to extract the text. First, I'm going to declare a variable IMG, and G equal to C two dot IMRT method, IMRaD. Then inside the roundresses, we need to pass the file location and our filename is simple dot JPG. After read the image, you can convert this image into grayscale image. Also, if you don't want to convert this image in grayscal image, yes, you can. But I would like to convert this image into grayscal image. I'm going to declare another variable. Gray, gray, equal to and from CV two, I want to run a method CVDlor CVT color inside the rounress first, we need to pass the variable in which variable we store the image, which is IMG, then we need to pass CV two dot color Bsa to gray, Colored underscore, BGR to gray. Now, let's show the image in a window. For that, I'm going to tie CV two dot, I am so method. Then the rounds. At first, we need to pass a window name which is simply D and then we need to pass the source image, which is gray. Next, I want to run wet key method, CV two wet key, I that the round brrass zero. So let's run the code and see is it convert this image gray scale or not. So to run this code, you make a terminal, as you can see, and it convert the colorful image into grayscale image. Now we need to extract the text from the image. For that, we need to run a method. We need to run a method from Pi resurrect module. So first, I'm going to do a variable text, equal to, I'm going to call this module Pi resurrect Pi resurrect from Pi resurrect, I'm going to run a function which is image to stream. So type image. Underscoe to underscoe stream. This one, the last one. Then inside the round presses, we need to pass the file path, the file source, which is gray variable. Now I'm going to comment out both the lines, I don't need these lines and I want to pin the text in my terminal. Some type print statement, print, Is the round dresses, I want to pass the text variable. Image to string method going to extract the text from this image and it's going to set this text in this variable. So let's set the file and run this code in my terminal and see is it to work properly or not. After run this code in my terminal, as you can see, it print the text successfully. This is simple text. Text is at different reason. If I open this image file, you can see both the text. I print as it is. It can identify capital letters, also it can identify small letters. Now we need to set this text in a document file. For that, I'm going to open a file. First, I'm going to declare a variable. File equal to open inside the roundresses. First, we need to declare a fine name, which is recognize dot PHD, and I want to open this file, W plus mode, so type W plus. Then we need to write the text in this file. So Five, dot, I'm going to call right function. Right. Then inside this write method, I'm going to pass the variable. I'm going to pass the variable, which is text. As you know, it is a good habit if you open any file, also you need to close the file. I'm going to type file dot close method. So this is it. If I set this file and run this code again, now you can see my terminal first it print, this is simple text. Then also, you can see my current working directory, it creates a recognized dot text file. If I open this file, here you will see the same text, which we extract from this image. This is how we can use Pi resact and OpenCV together, and we can easily extract text from an image. I hope now it's c for you, thanks for watching this video Stu for our next project. 24. Invisible T shirt using OpenCV: Hello, and welcome to another exciting Open CD project. Today, this project, we are going to create invisible T shirt. Let me show you the demonstration. As you can see, this is me and I am wearing black t shirt. And now I'm going to bring my invisible T shirt. As you can see, this is my invisible T shirt. If I wear this T shirt, I become invisible. And also you can see the bed all through this t shirt. It's become coopeti invisible. As like Harry Potter Invisible clock, we made our T shirt invisible. To create this kind of effect, I'm going to take red colored T shirt. Also you can take green color, otherwise blue colored, but red color is perfect for me. And in four step, I'm going to copy this project. At SAS, we need to store and capture the background frame. The, we need to detail the red color cloth using color detection and segmentation algorithm. Next, we need to generate mask and we need to substract the red color, and at last, we need to generate the final out that's when you create the magical effect. Without wasting your time, let's jump into the computer screen and start the protein. So finally, we are in my ser studio code editor. At first, we going to import the module. So type, import CV two. Then I'm going to import another module, and I'm going to import NumPi module. It is necessary. Import NumPi as NP. At first, we need to capture the video, and I'm going to capture the video using our webcam. Here I'm going to declare a variable name capture underscore video capture video, assign with, I'm going to call a function called video capture, Cv dot Video capture. We're already familiar with this function. Then inside the round verses, I'm going to pass zero. Now the question is why I pass zero? Because if we pass zero, then we can directly access our webcam. We can directly access our default webcam. If you have multiple webcam and suppose you have three webcam in your system, then according to your needs, you can change the number. I want to start my default web cam so I pass zero. Next, we need to provide a little time to our camera to arma. I'm going to use T module. He type, import, and I want to import time and time dot slip method. Inside the round bss, I'm going to pass one. Here I provide 1 second delay to our code. Then I'm going to declare variable. First, I'm going to declare a variable named count, count assign with zero. Then I want to declare another variable called background. Background assigned with zero a. Now we need to detect the background. Also, we need to capture the background. We can easily use this background for masking to capture it, we need to run a loop through this variable. So here I'm going to use for loop for I in range, and I want to set range up to 60. Then inside the for loop, I'm going to declare a variable, written val, return, underscore, value, val, comma. As I'm going to use this variable called background equal to I want to run capture dot Rt function. Capture dot Video Tat and I want to run read function. I want to read the video which we get from our webcam. So Tyt video, read read method. Then I'm going to set an IP condition. I written Vallow written Val equal to equal to false, then inside this IP condition, I want to continue this rule. Continue. It may pretty confusing for you. Let me explain this code step by step. This nippy Do Python code is used to capture the background frame from a video. In our case, we are going to read this video from webcam. Over a certain number of frame, in this case, 60 frame. Then it go to processing the video frames and extract a background frame for our further task. We can use this frame for background substraction, otherwise, special effect like green screen, et cetera. At first, this loop will run 60 time, which means the core inside the roof, we execute for the first 60 frame of the video. Then here we run a read method, capture video dot Red Method. It is a method from OpenCV, that's read a frame from video feed. And the first variable written well indicates whether the frame was successfully captured or not. It's written true if the frame is successfully read. Otherwise, it written falls if it fails. Then come another variable background. As you can see, first we declare background equal to zero. This background variable going to hold the actual frame which you grab from the video. Basically, it's going to capture an image, and then we compare a condition if written well equal to equal to false, this check whether the frame was successfully captured. If written well fails, the prographs skip to the next iteration of the room. And if the frame was not captured, the continuous state may tell the poorer to skip the rest of the loop code and process the next iteration t going to ensure that only successfully captured frames are considered. And now outside the four loop we need to flip the capture frame. Basically, we want to mirror the capture frame. Otherwise, it's going to return the opposite image. For that, I'm going to tie background equal to what we call a method from NumPi tie NP taught flip method. Then you said the run process first, we need to provide the source file, which is background. As you know, we stored the frame in our background variable. So to pass background, coma also we need to pass Xs x is equal to one. I going to flip the frame. Up to successfully flip the image frame, also, we need to flip the video frame. For that, here I'm going to run inside the roundreses captured video, and here I'm going to run is open method. Dot is open, then while loop, written al comma, image, heretical, another variable called image equal to capture video dot read Mnood capture video dot read after read the video, I want to reflect the whole video. I want to film the whole video. Then again, I'm going to run Hip condition. If not written well, then inside this IP condition, I want to break the loop, break and outside the Ip condition, I want to increase the count variable v. Count, count equal to count plus one because we need to incrementate the loo plus one, I want to increment this loop by one. So to read the whole video frame by frame, we need to. We need to flip all the frame. So to read all the video frame by frame, we need to flip the frames. Here, inside the loop, I'm going to type image, equal to. Again, I'm going to type np dot flip Method. Flip. Then inside the round presses I'm going to pass the variable IMG, comma X is equal to one. Then also interst the xs. X is equal to one. In our previous group, we read only 60 frame, and then I reflect one frame. But in this Valu, we read whole video and we select all the frames, and then we flip all the images. Here we flip our video and background image successfully, and then we have to convert the images into HSV color. By default, it's come with BGR color. Here I'm going to declare another variable called HSV. HSV equal to CV two to change the color, we need to use a method called CVD color, CVT color. Then inside the round press, first, we need to pass image, the source image, Summer type ING. Then to change the color, I'm going to call CV two dot color, underscore. From here, we need to select the exact vizier two HSV color. So type PGR two HSV. This one. It's going to change all the frame color into GR to HSV. I hope you are already familiar with HSV color, and now we need to create two mask for isoluting certain color from an image, especially we're focusing on red color too. Basically, we are focused on extracting red colored tone in an image. That has been converted to HSV color. Here, we need to create total two mask and each of the mask need two colored tone, lower red color tone and upper red colored tone. Here, I'm going to type a single en comin setting the lower and upper range for mass one. I'm going to declare a variable. At first, I want to set the lower red tone. This is for mass one. So who type lower underscore rate. Equal to here, I'm going to use Nb AA Mth Np AI. Then inside the roundss here I'm going to pass an array. Is the square passes, I'm going to type 140, 40. Also, I'm going to say the upper red wave. I do begin this line and I'm going to replace lower rate with upper red. And this time, I'm going to say it 100 value is 100, saturation value is 255 and value is 255. Now, let me explain this two line. As you can see, we create a number array and in this array, we have total three Vo, Henry p three Vo 140 and 40. The first value represent Hu, the second value represent saturation, and the last one is represent value. Basically, represent colored tag. I measure in degrees. For Open CB, we use 021 80 degree, but traditionally, its range is 023 60 degree. Then come saturation. Saturation determine the intensity or purity of the cool. Zero is gray and 255 is fully saturated color, and here we use 40 and the saturation we use relatively low. It means the color will be somewhat vast or fed up. And the last value control the brightness. It controls the brightness of the color. Zero represent black and 255 represent full brightness. The value we use is blue. Which means the color will be relatively dark. In this variable, we capture kind of greenish blue color rather than red color. At the same way, we say upper red Value. Also, upper red variable contain the same color, but its brightness and intensity is high. It is not fade out. And now I'm going to create the mask one variable. Ms one. This is mask one, Hemel mask one assigned with CV two dot, and I'm going to call a method called in range. I range. Then you said the round presses, I'm going to te HSV, comma will to pass the lower rid and the upper rate. Lower rate comma, also need to pass the upper rate. This is our first mask. Here we create a binary mask based on specified color range. As I told you, we need to create another mask. I do get this section and first, I'm going to change the command. This is for mask two, and also we need to use the same variable, lower rate and upper rate. But this time, I'm going to change the color angles. Also, I'm going to change the values. So here, I'm going to tie 155 degree. And for upper right, I'm going to pass 180 degree, and then I'm going to change the variable length. This is mask two. And now I'm going to combine both the mask. So here, I'm going to type mask one equal to mass, one plus mask two. If we march it, then it allow more comprehensive cotage of red colored detection. After march both the mask, we need to refine the mask. We need to remove all the noise from the mask. For that, hemorrhotide, mask one, mask one, assign W, here, we need to run a method from OpenCB which call morphology X. It can remove all the noise point from the mask. Hemorrhotide, CV two dot, morphology X. This one. Then inside the roundresses as first prater, we need to pass the source, the source image metal, which is mask one. After coma, here you need to call morphological opening operation. Again, I'm type C two dot morph open. This one, first we pass a binnymsk that was created earlier where white pixel represent the area of interest, and black pixel represent the background or ignored areas. Then we conduct an operation, morphological opining. It is a two step operation. First, it perform erosion, then it perform delusion. Erosion remove small white noises. Especially isolated pixels and dilution, restore the size of larger white areas. This operation is particularly useful for removing small objects. In our case, noise. Noise is not part of our mean detection. That's why we need to remove it. Now we need to create a structural element, otherwise kernel size. For that, I'm going to pass comma, and I'm going to use function from Numbi type NP Oce. Then inside the round presses, I'm going to pass kernel size three by 333 Club. Then also, we need to pass the data type of this pixel, which is NP dot UN tight. This unsigned Edwit indser is commonly used in Ig processing because it can represent pixel values. Then after coma, we need to pass the iteration value. Iterations, and I want to run it for twice. I'm going to pass two. This iteration specified how many times the opening operation is applied. In this case, this opening is applied twice. Multiple operation increase the tinning effect. That's why we need this iteration. After that, we need to perform the dilation in mask one. Again, I'm going to type mask one equal to here I'm going to call the method from CV two, which is known as dilate it. So type C two, dot, dilate. Then inside the roundss at first, we need to select the mask one. I'm type mask one, then we need to pass the kernel sides. Again type NP dot one. Inside the round brass, I'm going to pass three comma three. Then we need to mention the data dive, which is NP INT eight. Then atracma also we need to mention the iteration, but before, we need to move all this section inside the round ***. Here I'm going to mention the iteration. This time we want only one iteration. Now the quotien is what is dialed and basically what you going to do? This operation expand the white area of the mask, making the detected result larger and filling in small holes. And these s improve the mask robusts. Next, we need to invert the mask. We need to turn white pixel into black and black pixel into white. And to do that, we need to use VT is not operator. So first, we need to type mask, two to select the mask, two equal to Here type C V two, dot, biting, some type B, bitwise naught. We already learn bitwise operator. I'm not going to explain it here how bitwise operators work, like Bt nought, bitwise inside the Rundresses I'm going to pass mask one. Here we refine the mask corresponding to the detected red color. After refine our mask section, we need to generating the final out. And now we need to create total two separate output. In the first output, I'm going to show the background where red tlor is detected. In the second output, I want to exclude the ate areas. Let's create the first output for the background. Here, I'm going to type and I'm going to me this very well as resource one. R is one resource one equal two, and we need to use PTs Operator, PTI and operator. So tie CV two, dot, B wise, and then inside the round verses, I'm going to pass the background. The background is the image, which represent the scene where you want to show the parts of the images based on the detected red curve. First, we need to pass the background and we already create the background. Let me show you. This background. As you can see, we flip the background. I'm going to copy the background variable. I'm going to paste instant toys and operitoT is the background image, which represent the scene where you want to show parts of the image based on their detected red color. Then after coma we need to pass the source, source two, which is background. Next, we need to pass meta. And I'm going to type mass equal to mass mass one. Basically, these variables store the background where the red color is detected. Now we need to extract the current frames without red color. Again, I'm going to declare another variable RDs two. This is resos two equal to C two, again when we use tis operator. Is N. Then inside the round presses this time, I'm going to pass the video source. So as you can see, we set this video source as image variable, IMG. This is our flip video. I'm going to copy this variable, and I'm going to pass it here. Again, we need to pass the source two, this is source two, and I want to assign it with st two. I'm going to type metal mask, Tkoado mask two. Basically, B towise and operator perform operation on two images, otherwise, between an image and a mask. It is used to merge two images based on a logical operation that compares the pixel value at each corresponding position of the image. First source is first input image and second source is second input image, and the mask parameters specify which pixel to consider for the operation. Now we need to create the final o. So I'm going to declare a variable final underscore out. Final output equivalent to here I'm going to call a method, CV two, C two dot add weight. Sound type A, D, this one, aight it. Basically, we use this method to bind two images. Here I'm going to provide the first image, resource one, RD is one. Also, we need to pass the Alpha value, which is a float point. So to pass one comma then we need to pass source two, and this is R is resource two. And again, well to pass the float point. In this time, well to pass the beta float point. Again, I'm going to pass one. At last, I'm going to pass Gama Val. Gamma is zero. This function bind two images together by combining them with specify weight. At first, we need to pass the source one value, then we need to pass the Alpha value, then source two, then we need to pass the beta value, at last, we need to pass the Gamma value. Now we need to show the final output in a window. For that, I'm going to tie CV two dot, Im method, IMF method, IMF. Then the round dresses, I'm going to pass. First unit for byte the window name, which is invisible T shirt, and also we need to pass the final out. I copy this variable and apracma I'm going to pass the value. Final. Next, we need to exit the program where we praise a specified key. We prese ESC key in our keyboard, I want to quit the program. For that, I'm going to declare a variable K equal to CV two wet key method. Wet key inside the round presses, I'm going to pass ten millisecond delay. Then I'm set a deep condition I K equal to equal to 27, then in such a condition, I want to break the loop, break. I think I did some mistic. I did a very sllistic we need to move the code inside the indentation level, we need to move it intent six. I'm going to provide the space and move it inside the indentation level. Otherwise it's not going to run. One by one, I'm going to move all the line inside the intendion level. Now it's look perfect. As you know, 27 is asykey. It is code for skipped key. When we praise scape key, it's going to break the loop and it our program. Now, let's set the file and run the code in our terminal and see is it to work properly or not. So I'm going to hit Run button. After I hit the Run button, we need to wait for a couple of seconds. It takes time to start the webcam. Also, I'm going to move out in front of this webcam because we need to grab the background. As you can see on your screen, it open the webcam and also you can see, I am not in this window. So this is me and I'm wearing a black t shirt. Now I'm going to take my invisible t shirt. This is my invisible t shirt. As you can see it completely invisible, you can see the background world through it. Now I'm going to wear this t shirt and become invisible. After I wear this t shirt, as you can see, I'm become completely invisible. Also you can notice little glitches and is, but it is work pretty fine. This is the project where we build a invisible t shirt. I hope you like this project. Thank you very much and stay tuned for the next tutorial. 25. Background subtraction – OpenCV: Hello, guys. Good to see you back. Once again, I'm back with another project related Open CPI and today in this project, we are going to execute background subscription using Open SIP. Background subscription is a way of eliminating the background from image. To achieve this, we extract the moving foreground from the static background. Background substraction has several use cases in everyday line. It is being used for object segmentation, security enhancement, pedestrian tracking, counting the number of visitor, number of vehicle in traffic, et cetera. It is able to learn and identify the foreground mask. In Open Civi we have three algorithm to do this operation. Let me show you the algorithms. These are the three algorithms that we can use to subtract background. This is our first algorithm, background substract or MG. It is a Gaussian mixture based background foreground segmentation algorithm. We are going to use it first. At first, I'm going to import the libraries. I'm going to tie Import and our first libraries is NumPi NumPi as N P. Next, we need to import the most important library CV two. Tie Import. CR two. Now, first, we need to create an object using this algorithm. Which algorithm our first algorithm, data substrctorG? I to copy this algorithm name and first I'm going to declare a variable where I will stood our first object. Basically, it is a foreground to background substructor algorithm. I'm going to name it G BG, G VG one. This is found and vector subsector one, equal to hermano type, CV two dot, and we need to call another method, and our method name is BG AC GA, this one. Then I'm going to type dot, and we need to create the object type, create, and I'm going to paste our first background substructor, this one. At last going to pass down ss. This is the first object which I create using this algorithm, background substructor Mogen. Next, I'm going to capture frame from the camera. So I'm going to take a variable CAP capture, CAP cap, assign with her and call C dot videocapture Metal. Video capture. Then inside the round recess, we need to pass the sources. We are going to turn on my default web cam so I'm going to pass zero. After that, here we need to run a infrate while loop. So I'm going to type I inside the roundres one, I mean i true. If the condition is true, then inside the i, first, I want to capture the frame. We need to read frames. To read frames, also need to declare Gulen value type. This is the Willian Valu stored to Bullen Value also I am going to return the image from the frame. I'm going to type ING, equal to, and we need to run the read function. Cab dot read. This method will read on the frames. At first, it's going to return a vol value, what a Bullen value. It going to store if there is a frame or not. Then then the image, it going to store the actual image, otherwise, actual frame. Then inside the VIL, going to create the mask, the first mask using this image. For that, I'm going to create a fun mass G mass one. This is G mass one, equal to Helm to apply this image to this object. Helmer type G, BG one taught apply mass boot. Then you said the round presses, I'm going to pass IMG image. Here, we apply mask for background substruction. Then I want to display this mask and also I want to display the original wing. At first, I'm going to display the original window. I'm going to tie C two dot IMF metal, IM so then inside the round resist inside the single coat, this is the original window, where I want to do fi webcam footage, some type original and here I'm going to pass variable image. As you know, image variable stored the original footage. That's why I need to pass image. And then again, I'm going to call Ema Show math. I duplicate this line. But this time, I want to show the Mg algorithm. I copy MG and plus name, replace original with MG. I replace IMG with GmGmsk G mask O. And now we need to call wet K method. Here tie K equal to CV two dot wet K method. Then inside the round reverses, I'm going to pass 30 millisecond. Also, here, I'm going to pass BTIs operation, so I'm going to use operator and I'm going to pass a By code, zero XF. Then we read this line. Then I'm going to set a condition for SkepKHertype, I K equal to equal to 27, then inside this E block, I want to break the loop. Break. As you know, 27 is the code for scape key. Then after break the loop, I will set the IE loop, I want to release the camera. So tied, capture, CAP cas, tot, release meth wood. Rise. Also, I need to destroy all the windows, so I will type C two, tot destroy all the window. So let's set the file and I'm going to run it. After I run this code, it's going to open total two window. In our first window, it going to the original video footage. I want to see the colorful video footage and the second window, it going for the for down background mask. Now, let's set the code and see what it going to return. I want to hit the run button and we need to wait for some time to start the camera. As you can see, turn on my webcam and I'm going to move this window beside of this original window that you can easily compare. As you can see, this is our original window and this is the MOG mask. As you can see this filter written bitwise operator, and when I move my hand, otherwise body, you can identify the moving portion with white color. Basically, it defined the area with white color. This is the moving part of my body. That's why it captured these sections. And if you want to become invisible, just stay still and don't move. As you can see time to time, I become invisible and when I move my hand, then again, it track the moving area. This is how we can substract background and display only the moving portion. Now let's talk about the next algorithm. I'm going to delete on the windows. Now, let's create the next object using Mg two algorithm. Background substructed Mg two. First, I'm going to duplicate this line and I'm going to change the variable name. This is MGMG two. Also we can remove it. We don't need this method for this, and I'm going to replace MOG with Mog two. Then I'm going to create a mask. For that, I'm going to duplicate this section and replace mask one with mask two and also we need to replace the variable, this one, Healtide GBG two, and we apply this image to this second object. After that, we need to display it in our window. I'm going to duplicate this section and I'm going to display mask two. I'm going to change the window name. This is MOG two. Let's see after I apply this function, what it going to return. To set this file, again, I'm going to run this code in my terminal, and we need to wait for a couple of seconds to start the webcam. Now, as you can see, we have to tell three Window. This is MOG two, this is MOG and this is the original Window. Now let's see the difference. As you can see, if we compare Mg two with MG one, OG two provide better adaptability to varying scene due illumination changes. I want to say it adopt light more accurate than MOG. That's why you can recognize my facial expressions. As you can see, when I move my hand, Mg two return the complete image of hand because this algorithm read lights better than MOG. Now let's apply our next algorithm. I will do only wins. This type, I'm going to apply better substructor GMG. Again, I'm going to duplicate this section and replace GBG two with GBG three. This is Fun background three. Again, we need to type Cv dot BG ACG, create Bagnew substructor and here we need to pass GnG. And here you can see this algorithm combine statistical background image, estimation, and perfixel basin segmentation. Bgrou substructor Mg two is also a Gaussian mixture based background foreground segmentation algorithm. Let's create a mask for the third algorithm. Super create the object, I'm going to create the mask. I'm going to create this line. And I'm going to replace mask two with marks three and also need to replace the object name. And also need to change the object name, GBG three. After I create the mask, also we need to display this mask in windows. Again, I'm going to duplicate this line and replace Mg two with GnG here, we need to display the mask three for down background masks three and we'll do satisfied. After step this file, again, I'm going to run the score in my turning. Now you can understand all of the algorithms. Now you can see we have total four window and I'm going to arrange it according to the order, original OMG, OMG two, and GMG. This is our original window. This is the MG window, and this is Mg two and last one is G&G. Now you can identify the difference between all of the. You can compare all of three result with each other. If you notice GMG here you can notice once the background model is established, the algorithm applies basin segmentation at each pixel and Bayesian segmentation is a probabilistic method. That calculates the likelihood of a pixel belonging to the foreground or background based on its intensity. Also you can notice it exaggerate the pixel of orders. But all of these three algorithms work perfectly. It substract the background and represent it with black color and display the moving portion with white color. That's why these algorithms are widely used in real time video analysis, surveillance camera, and traffic monitoring system. Also, it is used for object tracking. So this is how we can substract background from foreground. I hope now it's clear for you. So in this tutorial, I show you, we have total three algorithm that can be used for background substriction. And according to me, Sig one is best MOG two. Otherwise, you can go with MOG. GG algorithm is pretty noise. These algorithms are pre built in Open Series, so we don't need to do too much for that, we need to apply it. I hope you like this project. Thanks for watching this video, stay tuned for the next tutorial. 26. Face Detection using Python and OpenCV with webcam: Hello, it's good to see you back. Once again, I am back with a new project related Python OpenCV. And today in this project, we are going to create phase detection. Yes, we are going to create phase detection using open C. So before I start our program, let's see the demonstration. And as you can see, it successfully detect our face and create rectangle around it. And whenever I move my face with my face, it's also move the rectangle. It's mean our program work perfectly. It can detect faces. And if I move my hand in front of my face, now you can see it removed the rectangle and if I move out it, again, it show the rectangle. So without wasting your time, let's start the practical. This project utilized OpenCV library to make a real time phase detection using rob Cam as a primary camera. But to create the project, we need to download something. We need to download a hear casket classifier fine. So let's jump into the browser. As you know in this project, we are conducting phase detection. So here I open my search engine, Google, and here I search har Cascade, nscFontPas underscodfault, dot N. Yes, we need to download a EML file. After search this file, as you can see, it provides a website, GitHub link, arcascadeFontalfas, underscofat dot N. I'm going to click on this link. I'm going to redirect to this page, and this is the ML page we need to down. There are lots of hard casket find available for different body part, such as eye, eyeglasses, fontalfas, fontalpas default, full body, lip type, and many more. But we need to select this one, Hardcasket underscoFontal phase, undersco default. A the big question is, what is that and why we need to download this file. This file is a pre tent classifier used by Popsii for detecting phases in images. This classifier is trained to detect human fontal phases. A HRGske is an object detection algorithm used to identify objects in an image or video stream. It works by training the model to recognize pattern in the image using her like features, which are similar to edge dection filters. And this classifier is already built by training on a lot of positive images such as image of fontal face and also negative images. I want to say image without faces. Based on this training, the classifier learns to detect facial patterns. So already saved a lot of facials algorithm on this file. So we don't need to teach our model by ourselves. I'm going to download this file and I'm going to wp this file in my current working directory. So as you can see in my current warranting directory, there is a file name hung cascade, underscoeFondl phase underscoe default dot L we need this file. It is a pre trained har cascade classifier, which is a specify algorithm designed to detect phase in images. In a single sentence, this file contain the necessary data to detect phases in images. At first, we need to import the library. Here I'm going to import our library OPC. Import CV two, and then we need to load the pre tad hard casket fine. For that, I'm going to declare a variable and I'm going to declare and our variable is A, A, assign Wi, CV two dot from here, I'm going to call the meth cascade classifier, some type cascade classifier. Then inside the round brass, inside the double codes, I'm going to copy the file path. Otherwise, define herame. Here, I'm going to copy the fine name and back to the main dot Pi file. Then inside the double code, I'm going to paste the fine name, HcascdUnderscoe fontal phase, underscoe default, XML. This file is saved in XML file format. Next, we need to capture the video feed from the U webcam. For that, I'm going to declare another variable, and our variable name is CAN Cam equal to from here, we are going to run a Math mode, CV two dot video capture. Then inside the round recesses, we need to provide the camera source. Here I am use my default webcam, so I'm going to pass Z Next, we need to run an infinite loop to capture the frame continuously from this camera. For that, we are going to run a uh in inside this IU, we need to read the frame from the camera. Hem type Cam Cam dot read Method. Read this method, going to return to dull two vdu. So we need to store this vario in a variable. First, it return a Boolean value that indicating the frame was successfully read or not. So I'm going to sort this Bolen value in RET variable. C then it's going to return the actual image. To store the image, here I'm going to declare another variable, CNKscimage IMG, assigned with Cam dot read function. This variable going to store the boolean value and this variable going to store the actual image, which you get from this frame. And now we need to convert this colorful frame into briske image. Soap hair, I'm going to type gray. Gray, assign with, I'm going to run a function called CVD color. Cv CVT color. Then inside the round res as source, I'm going to pass this variable cab image. We are going to convert the CaM image files into gray color. Then as second source, we need to pass the initialized code, which is CV two dot, and I want to convert color to BGR color to gray colored. So type, colored under scope and we need to convert BGR color to gray colored type BGR to gray color. This one. Basically, phase detection work better on grey scale images, which simplify the process. Now we need to detecting the phases from the grey scale frame. For that, we need to use a meth detect multiscale. Here I'm going to declare another variable and my variable name is phase, assign with a dot, dettT one. Then inside the round presses, first, we need to pass the source image which is great. Then we need to pass the scale factor, which is 1.3. Next, we need to pass minimum number of never rectangle. For that, I'm going to pass six. In this method, first we pass the image source gray. Then we pass the scale factor win. Here I pass 0.3. 0.3 is the scale factor that control how much the image size is reduced at each image scale and six is the minimum number of levers. A rectangle should have to be considered as a phase. This phase variable going to store all the phases from the grayscale image and except the phase coordinates. Now, using these coordinates, we need to draw rectangle around it. So let's draw rectangle. We need to draw rectangle around the detect faces. For that, we need to use form. For in at the Y. So I'm to type four inside the roundresses here, we introdu some coordinates as we do X one, Y one, W one, H one in phase. Then inside this pl boop, we need to draw rectangle using this coordinates. So as you know, to draw rectangle, we introduceectangle method. CV two dot rectangle. Then you set the roundressF we need to pass the source image in which image we want to draw the rectangle as source image I would pass this variable CaM image. Ma. Next, we need to pass the points to draw the rectangle. But before we need to understand the points. This four loop iterate overall detective phases which is stored in phase variable. And for each detected phase, XY and Y one represent the top lip corner of the phase, and W one and H one represent weed and height. These two value combined represent top lip corner and this value represent wet and this value represent height. Using this to using these coordinates, we are going to draw the rectangle. And as you know, to draw the rectangle in OpenCV, we need to pass total two, top lip corner and bottom right corner. For the top left corner, we already have the coordinates. So here, I'm going to pass X Conga Y, X one, Y one. Then then we need to pass the bottom right corner value and we don't have the bottom right corner value, but we can calculate the bottom right corner value. Let me show you how. To calculate the bottom right corner value, we need to addition with y1s value with rectangle with, which is one. Then we need to calculate YR plus H one, height. At first, we addition with with xs value, then we addition height with Y xs in. Combined, it's going to return the right button corner v. Now, we need to pass another value which is colored here I'm going to use blue colored. First, I'm going to pass 2550, zero. Next, I'm going to pass the ness, the thickness of the rectangle border. I'm going to pass three and also you can use higher value for that if you pass five or six. I'm going to pass five. After I draw the rectangle around the face, we need to display the frame with the detected phase. To display the frames at detected phase, we need to go outside from this four loop and here we need to display IMF metal. C two dot Imho they said the roundresses. At first, we need to provide a window name and window name is face, and then we need to provide the source image, which is Cam Image. That's. Also, we need to set the wet key to stop the program. Here, I'm going to type I CV two dot weit key inside the roundre I'm going to set it for one again, one milliseconds and zero F equals to equals to ORD. Inside the run presses, if we praise small A, then inside this p condition, I want to break the loop. I want to break the Y loop, break. If we praise A, then it's going to bring the loop and kill you into. Next, we need to release the web cam resurce. Outside the ID loop, Hemel type, Cam dot, release method. Also we need to destroy all the window, S type, CV two, dot, destroy all Windows. This one. I'm going to set this file. Now, let's run the code in my terminal and see is it to work properly or not. After set this file, I'm going to hit the Run button, and we need to wait for some time to start my webcam. Finally after wait, couple of second, it open our window. As you can see it successfully detect our face and create rectangle around it. Whenever I move my face with my face, it's also move the rectangle. It's mean our program work perfectly. It can detect faces. I I move my hand in front of my face, now you can see it removed the rectangle and if I move out it again it show the rectangle. Using this pretend fontalPashar casket file, we create this detector. Now let's close the window and back to the visual studio code. So I successfully create phase detection program using OpenCV. It is not a very difficult one. It's a very easy project. I hope you enjoy this project very much and also understand how phase detection work. So thanks for watching this video schedule for our next project. 27. Detecting objects of similar color in Python using OpenCV: Hello, guys, good to see you back. Once again, I am back with a new project related Pythonopen CV. Today in this project, we are going to perform object tracking using color segmentation. Yes, we are going to track object using color based segmentation. Object tracking is a computer vision technique for locating position of an object on image or video. Remember, recognizing the object and tracking the object are two different. In object tracking, you can track any moving object, but in object detection, you can identify the object. These are two totally different things. There are many type of object tracking method in OpenCV, such as color segmentation based, frame differencing base, and feature matching base. I institutorial, we are going to perform color segmentation based object tracking, and to perform the effective color tracking, we are going to convert BGiR color into HSV color. As you know, HSV stands for hue saturation vue. Hue stands for color, Vallow stands for amount of color and saturation stands for gray amount of color. Without talking too much, let me show you an example what we are going to create in this tutorial. Now you can see it successfully track our object. It can track our blue colored bottle cap. In our frame window, you can see the original footage and in our mask window, here you can see the BTIs operator footage. Here, white represent the bottle cap and black color represent unselected area. But in our third window RIAs, here you can see the actual color object. Here you can see the exact bottle cap. Without this bottle cap, you can't see anything here because it track only this color segmentation from this window. This is what we are going to create in this project. Let's jump into the computer screen and start the project. So as you can see, finally, we are in my visa studio code editor, and I already created a Python file main dot py. At fast, I'm going to import our important library. I'm going to type import, and I'm going to import CV two. Then I'm going to import Numpi. Yes, Numpi is important to create this kind of project. I'm going to type import NumPi as NP, next, we need to import Pi plot from the matplot leaf. He type, import matplot leaf Pi plot as PLT. Here we are going to perform object tracking based on the object color. Object tracking and object recognizing is totally different things. In object tracking, we can track any moving object. But in object recognizing, we need to recognize the particular object. As I told you earlier, there are many methods that we can use for object track, such as color segmentation, frame differencing, feature matching, machine learning waste, et cetera. But as you know, we are going to perform color segmentation based object tracking. Now let's start the practical. At first, we need to create a video object to capture the frame. To capture the frame, we are going to use our webcam. Here I'm going to type a variable name, CAP, CAP, Assign W and here we need to use a method called video capture. Cv dot Video captured this one. Then inside the round dresses, here we need to pass the source Z. Then after that, I'm going to define a empty function. Why we need empty function? Because we are going to create track bar using it and to create this track bar, we need to return empty function. Who is going to return nothing? Here, I'm going to type Dave and our function name is nothing. Nothing. Inside the round dresses, I'm going to pass a normal parameter called X and inside the function, Hemo type pass. Basically it's going to return nothing. Then using this empty function, we are going to create track bar. But before I create track bar, we need to define the track Bar name. Here, I'm going to type V two dot named Window. Here I'm going to use a method called name Window named Window, and we need to provide a name for this window, for our Tug bar Window I'm going to type tracking now here, I'm going to create six track bars because we need to grab upper Hue saturation value and we also to grab lower Hue saturation value. Let's create our first track bar. Here I'm going to type C two dot, create track bar. This function. Then inside the round brasses, at first, we need to create track bar for lower bound value for Hu. Here I'm going to use the short from lower Hu H. After you create the track bar name, we need to pass the window name. In which window, we want to show the track bar. Inside the double codes, I'm going to pass this window name tracking. Next, we need to pass a value. Here, we need to pass the lower value, which is called zero. Then we need to pass the higher value and for higher value, I'm going to pass 255. At last, we need to pass a function. This function will return nothing, so I'm going to use this function, nothing. Just want to copy the function name and put it here. I'm not going to set any default value to our track bar because we are going to set it minually. Then I'm going to duplicate this line and this time I'm going to change the name of this track bar, which is As. As stands for lower saturation. Then I duplicate it A and this time I'm going to name it L V, lower value. These three track bars stands for lower bound value. Now we need to pass upper bound value. I'm going to duplicate the whole three line. So our last three trig word is for higher value, higher hue and syn is higher saturation and last one is higher wave. Now we need to run infinite loop to continuously process the video frame. For that, we are going to use While loop. Here I'm going to type YT then inside this illoop I'm going to use read method. Read method capture each frame from the webcam to store the capture value and the Bullen value, I'm going to declare total two variable. First, I'm going to declare B, which is going to store the Boolean value, then I'm going to declare another value called frame, frame equal to cap dot read method. Then after read all the color frame, we need to convert color frame into HSV. As you know, by default, OpenCBuse BGR color format. We need to convert it into HSB color format. I'm going to declare it very well Hs V equal to V two, dot, CVD color. Then inside the rounddresses, I'm going to pass the frame. At first, we need to pass the image source. I'm going to pass the frame variable frame. Then I'm going to convert it CV two dot, color, underscore, and I'm going to convert it BG HSV. Search for GR to HSV. This one. Next, we need to get HSV value from these track bars. For that, we need to declare different variables to store the track bar value. So for lower, I'm going to type underscoe H. Equal to grab the data from this track bar, we need to use a method called gate track bar pause. I'm going to tie CV two dot gate, track bar, pause, this one, and then inside the round presses, inside the double codes, first, we need to pass the track ban name. From which track bar, I want to grab the, which is H, H. Next, we need to define the Window name, and our window name is tracking. I'm going to use this window name, tracking. I copy the window name and inside the double codes, I'm going to paste it here. At the same way, we need to create five other variable to get the value from the track bar. I duplicate this line. This time, I'm going to extract value of AS. He type underscore is also interchange the track banname L is. At the same way, I'm going to create all of it. This one is for V, and I'm going to change the name, V. Then I duplicate this three line and I'm going to make this one HH H underscoeH. This is for high U value. Next one is for high saturation value, HH the last one is for HV. Also to change the track but name, which is H V. So these lines retreat the current position of the HSV track bars, which represent the lower and upper bound of the HSV color range. We are going to use it for masking. Next, we need to create lower and upper HSV bound array. To create those array, I'm going to declare a variable L underscore B. It stands for lower bound y to create this lower bound array, we are going to use NumPi p dot array, Is the rounds, here I'm going to pass all the lower bound value Lunde H, underscore a, and underscore V. At the same way, we need to create upper bound value. I duplicate this line and I'm going to change the variable name. U UV stands for upper bound value. Here also interchange the variables name. I'm going to make it HH. HH stands for higher value Hs HH stands for higher saturation, higher U, higher saturation and higher value. Next, we need to create mask using lower bound value and upper bound value. To create this mask, we are going to use in range method. This is the most important part of this video. At first, I'm going to declare a variable called mask mask equal to CV two.in range math boot in range. Then inside the round presses, first, we need to pass the source. As source, I'm going to pass this variable HSV. So tie HsV. Then we need to pass the lower bound value. As you can see, we store the lower bound value using this variable, underscoe B, lower bound. L underscoe B. Also, we need to pass the upper bound value. And we represent perbund value using using underscore, B, that's it. Here we create mask using ne function. In function used to create mask based on color values in the HSV color space. At first, we pass HSV color space. This is HSV color space image. We convert BGR image into HSV here. I pass this variable. First we pass array of lower bound value, then we pass array of upper bound. Now, the bit inne function convert these values into a binary image, and we stored this binary image into mas variable. As you know, binary image create it total two value. While zero represent black and 255 and 255 represent white. Now this mask highlight the reason in the frame that match the specified color range. Suppose you select blue colored innu track mark, then it's going to show only the blue colored portion in into. Now we need to apply this mask with BtIs and operator. Here, I'm not going to explain in depth how Boys and operator work. To apply the mask on Bitwise and operator, I'm going to declare a variable RES. Rs equal to, we need to use BTI and sub type CV two dot B twice and operate. Then inside the round brasses, at first, we need to pass the first source, which is frame as second source, I'm going to pass another variable frame. In our first video, we pass the original frame and in our second video also we pass the original frame because since we are masking the original width itself, and now we need to pass the mask variable. Hemotype mask equal to the mask. This create an output that shows only the part of original frame that match the specified color range. Basically, you're going to isolate the object based on the color. Now I'm going to display the video and the mask output, also the ras variable. Here I'm going to use IMso Mathod some type V two dot IM show. Then inside the round presses, first, I'm going to pass the window name in which window I want to display the frame. Our window name is frame. Then I'm going to pass the variable frame. Basically, this window going to show us the original footage, the colorful footage, Bier color footage. Then I'm going to show you the mask footage. I duplicate this line and replace frame with mask. And at last, I'm going to display this res variable, this footage, this up tar intersect with Bt operator. So here first, I'm going to change the window name Rs. Also, I'm going to replace this variable mask with REs and I'm going to set this file. Now we need to set the IP condition to break the loop using Wetey method. Here I'm going to type E condition I CV two dot wet key wet key. Inside the round process, I'm going to wait for 25 millisecond and zero x FF equals two equals to ORD. Inside the rounder says, if I praise small Q in our keyboard, then I want to break the loop. Break. This is it. After that, we need to release the capture. Cap release method. This is going to release our video capture. Also to destroy all the windows. V two, dot, destroy all windows. So we are done, but here I did a silly mistake. As you can see, it's an array, so we need to pass this video inside the square basis. Here at square basis. Now I'm going to set this file again and I'm going to run this code. Now, let's run the code in our terminal. After a click Run button, we need to wait for some time to start our webcam. As you can see, her need to open our window. Here we have to tell four window. First one is for tracking bar, and second one is for original frame and third one is for mask. It is for Bis mask and I'm going to place all the windows side by side that we can compare the result very easily. As you can see, this is the single colored bottle cap that I'm going to use to par from the object tracking. Here I use solid blue colour variant bottle cap. You can use any solid colored object, and I'm going to match this color using our track bar. And now we need to play with this track bar to match the bottle cap color. We need to play with the six value. At first, I'm going to increase the high value. Then I'm going to increase the high saturation value. We need to adjust these bars to get the perfect result. As you notice, when I change the values, it reduce the noise and try to focus on our bottle cap. I'm going to first far this process to get the perfect result. Now these coordinates nearly give me the perfect result. Now you can see it successfully track our object. It can track our blue colored bottle cap. In our frame window, you can see the original footage and in our mask window, here you can see the BTs operator footage. Here, white represent the bottle cap and black color represent unselected area. But in our third window REAs, here you can see the actual color object. Here you can see the exact bottle cap. Without this bottle cap, you can't see anything here because it track only the color segmentation from this window. Now let's kill the window and back to the program. This is how we can perform object tracking based on color segmentation. I hope now it's clear for you how we can create it following the nearly same process, we can create air Canvas with it. We're going to learn about it in our upcoming tutorial. Don't worry about it. Thanks for watching this video. Stay tuned for our next tutorial. 28. Import modules and setup mediapipe object: Good to see you back, guys. Once again, I am back with a new exciting project related OpenCV. And today in this project, we are going to create Finger Counter, with the help of Media Pipe module. Now, let me show you an example. One we are going to create. So as you can see, it start our web camera. And now I'm going to raise my closed hand. As you can see, now it show zero fingers and also it track all the fingers successfully. Here you can see all the landmarks, and now I'm going to spread all of my finger. As you can see on your screen, now it print five finger. You can notice it track all the finger joints and teeth. And if I close my thumb, now you can see it print four finger. And if I rise it again, again, it print five. And if I move my index finger, as you can see, now it print four finger. At the same way it track all the fingers. It count all the fingers and also it track all the finger coordinates. This is how our finger counter work. Finally, we are in my Visual Studio code editor, and as you can see, here I already create a Python file main dot pi. At first, I'm going to import our main module, CV two. So to type, import CV two. Then we need to install a important module, which is media pipe. To install the media Pipe, we need to type this command. P install OpenCb Python Media Pipe then you need to hit Enter. As you can see, requirement already satisfied. I already download this module, but you need to complete your download process. Then back to the main dot Pi Pi and this time, you need to import media pipe. The media pipe module. So to type Import Media pipe as MP. Now you might have question, what is Media pipe? Media pipe is a powerful open source framework by Google, designed for building multimodel machine learning pipeline. It provides a collection of cross platform library and pre trained model for implementing computer vision and machine learning task in real time. And if I talk about its common usage, it is used for phase detection. It can detect phase and track phases in real time. Then we can use it for pose estimation. It can identify human body pose, which is useful for fitness app. Also, we can use it for hand tracking. Using it, we can track hand movements and recognize the gesture. Also, we can use it for three D object detection for argumented reality. But in this tutorial, we are going to use hand tracking. This module is very lightweight and fast. Also, it is work on variety of devices, including low power device. Now, after input our libraries, we need to initialize the media pipe hands. For that, first, I'm going to declare a variable and our variable name is MP, underscopean, media pp hand, media pipe hand equal to PTT, solution from here, I want to call hands. It's going to exit the hands solution from the media bith then hands equal to punch P undercoe hand that I'm going to run a method called hands. Then inside the rounds, here we need to pass some parameters such as static image mode, maximum number of hands, and minimum detection confidence. First, I'm going to pass static image hand, underscore, image, underscore, mode, static image mode equal to, and I'm going to make it false. Then I'm going to pass maximum number hands. Maximum max, undercoe number, NUM, underscoe hands, maximum number of hands, equal to, for now, I'm going to pass one. Don't worry. I'm going to explain it later. At last, I'm going to pass minimum detection confidence. So type, mean, underscore, detection underscore confidence, equal to hermon to set it 0.7. As you can see, here we run a method called hands, and as a first parameter, we pass static image mode. We set it falls. What does that mean? It's mean it is optimized for video, and it detects hand only in the first frame and tracks in subsequent frame. Then at the second parameter, we set the limit. Here we set number of hands limit, and here we pass one. He limit detection to one hand. At last, as you can see, herro use another parameter, minimum detection confidence. Here we set minimum detection confidence zero point side Is wing it sets the threshold for the detection confidence. Lower thresholds may result in false positives. Next, we need to define the MP drawing. It is a utility for drawing landmark and connection on detected hands. Let me show you. Here type, P underscoe drawing equal to hero type pt, a method called solution dot drawing Utes, underscoe utils. As I told you, it is a utility for drawing landmarks and connection on detected hands. So basically, in that section, Hero initialize media pipe hand. Next, we need to create a function. We need to create finger counting function, and I'm not going to do that in this video. So this is it for this video. At the next video, we are going to define a function. So thanks for watching this video, Sta tune for our next studio. 29. Creating finger counter function: Good to see you guys. This is another part of this project. And here we are going to define a function that can count our finger. Also, it can count our thumbs. In our previous tutorial, we initialize the media Pipe hand, and here we are going to create a function. We're going to define a function called D Dave our function name is count finger. UnderscoeFinger. Count finger, and inside the surround recess is going to take one argument, which is hand landmark, underscoe landmark. Here we are going to pass the detected hand landmarks to this function. Then inside this function, I'm going to define a list, a empty list, and our list name is fingered. Fingers equal to a empty list. This list going to store the state of each finger. And now we need to define the landmark. For that, I'm going to declare a very well call landmarks, Landmarks equal to hand, Landmarks Landmark. Landmark is a list of 21 points, W going to hold X coordinate Y coordinate and z coordinates? It is a list of 21 points corresponding to different parts of the hand. As you can see in this image, these are the Landmark coordinates, which is provided by Media Pipe. This is how media pipe read our hands. As you can see, this is our thumb finger, this is index finger, this is middle finger. This is ring finger, and this is pinky finger. As you can see, it can detect all the joints of our finger. Also, it can detect the tip of this finger and each of the coordinate defined with a number. We are going to use these numbers and XXs and YXS coordinate to count the fingers from our hand. Let us back to the is studio code. So it can recognize 21 points from our hand. Now we need to detect the thumb. We need to check X coordinates for left hand and right hand orientation. For that, HemrotypFingers, dot append method. Inside the round dresses, hemotyp landmarks and I want to is from this landmark, I want to is fourth index, Tt, I want to excess Xs coordinates and less than Landmarks inside the square ss Amel pass index three dot XX. As I told you, our hand divided into 21 landmarks. As I told you, landmark is a list of 21 points each point represents specific part of our hand and each point hole total three coordinate. X xs coordinate, XS codent and Z xs coordinate, and Landmark index four represent the tip of the thumb as per media pipe, and Landmark three represent the thumb joint near the tip these are the X coordinate of the thumb tip and the joint. Now let's talk about what is the logic behind it? Why I use smaller than sine? Why Landmark four dot x less than Landmark three dot X? It's mean if the coordinates of the thumb tip is smaller than coordinate of the joint, it means the thumb is extended outward. This is usually true for the left hand when the thumb is extended and for the right hand, this logic works because the frame is flipped horizontally. And it ensured that the X coordinates are constant for the detection. Now after I understand the logic, we might have questioned why we need to append this in this finger list. Here, finger dot apens store the result, true either falls. If the condition is true, its main thumb is extended outward. If fails, thumb is not extended. So this code is for thumb detection. Now we need to detect other fingers. For that, I'm going to use four loop for Pep in inside the square ss, I'm going to check Y coordinates for that 812, 16, 20. Then inside this four loop, now we need to understand the four loop. What does that mean? For tip in eight, 12, 16, and 20. It's mean it iterate through the tip of index finger, which is represented eight middle finger tip, which is represented with 12, then ring finger tip, which is represented with 16 and pingy finger tip, which is represent with 20. Then inside this four loop, Hematite fingers dot append inside the round resis, mark inside the square resis, P, and I want to access the Y Y coordinates. Less than Landmark, inside the square resis, PP minus two, I want to access the Y coordinates. Now the question is, what is the meaning of this condition? What is the meaning of this logic? Basically, it's going to check our fore finger are extended or not. And Landmar TP, refer the current finger tip. Landmar tip dot YX, refer the current finger tip and Landmark tip minus two dot YXsRfer the base joint of the same finger. Let's try to understand with a real image. As you can see, this is the way media pipe observe our hand. As you can see, eight is the index finger tif and 12 is the middle finger tip, and 16 is the ring finger tif and 20 is the pinky finger tip. And if we minus two from the tip point, then you can see the joint. Here you can see the base joint. If we minus two from the eight, it's going to return six. Then six is the base joint of this index finger. At the same way, we can identify other finger based joint, ten, 14, 18. So this is Y coordinates of the tip, and this is Y coordinates of the base joint in our fingers. And here we compare the Y coordinates of the tip with the base joint. And if the tip is higher, the finger is considered extended. And if the tip is lower, then finger is considered folded. And our append method going to add the true false result for each finger in this finger list in this empty finger list. If the condition is true, it's mean the finger is extended, and if it falls, it's mean the finger is folded. So now we have the list that can identify the extended finger and folded finger. And now we need to count the list. For that, outside this four loop, her type return, we need to count the finger list, nothing else. Finger, fingers dot, count method, count. And inside the runbs we need to count only the true values, not the false values. So here I type, true. That's it. Now, let's wrap up this section E. And side by side, as you can see, I open my visual studio code editor and this image to understand the diates the landmarks. Here, first, we define a function and our function name is count finger, and it's going to take hand landmark as argument. Then here we create a EmptyList list of fingers. It's going to store the value in two or false? True means the finger is extended and false means the finger is not extended. Then we access the landmark. Landmark is a list of 21 points, and as you can see, these are points provided by media Pi and each of the point hold total three coordinates. XXs YXs and ZXs. At the next part, we dett the thumb using this append method. As you can see, this is the thumb image of right hand. And this is the landmark number of the thumb, which is four. And from this thumb, here we extract the xs of this thumb value. Then we access the thumb joint, which is landmark three, and from this landmark three, again, we extract the xs value. And if the condition is true, then it's mean our thumb is extended. If Landmark four S value is lower than Landmark three x value, it means the condition is true. It means the thumb is extended. If you notice you can see the distance between the borderline from this Landmark four is something 20 pixel and the distance between borderline two, Landmark three is something 60 pixel or 80 pixel Landmark three X value is higher than Landmark four, I mean the thumb is extended. If we fold our thumb, ten, landmark four will would be greater than Landmark three. I mean the thumb is folded. Here we compare thumb tip with thumb based joint. At the same way for other fingers, at first, using this loop, we access all the tips. In this list, here we pass index finger tip, middle finger tip, ring finger tip, and pinky finger tip. Then we extract the Y coordinates from this tip, and then we compare it with Ys direction from this way. For our thumb, we compare with Xs direction. But this time we compare with YXs direction. If the thumb base joined value greater than thumb tap, then the condition is true. And it's going to append the true value in this list. And next we count the true values. How many true value we have in this single list? Suppose we have total five true value in this list according to this image. Then it's going to return five. It's going to count all the true values. If one of the condition is false, then it's not going to count the false value, then it's going to return four. So this is how our function going to work. So this is it for this video. At the next part, we are going to start our main loop. We are going to read frame. So thanks for watching this video, Stay tuned for our next part. 30. Creating the main loop and run the project: Good to see you guys. Once again, I'm back with another video related to this project. And in this video, we are going to learn the main loop. At first, I'm going to open the webcam. To open the webcam, first, I'm going to declare a variable, and our variable name is CAP, CAP equal to cv dot Video capture Method. Video capture. Then inside the round presses, you need to pass the camera source. I'm going to turn on my default camera, so I'm going to pass zero. Here I'm going to type Y, cap is open, is open. Then inside this while oop, first, I'm going to run read method. Cap dot read. As you know, read method going to return to two result, a Bullen value and the original frame, and to store it in a variable hero type to store the wollen I'm type success. Comma and to store the frame multi frame, equal to Cabot read method. Then a mutua on a EP condition. If not success, then inside the IP condition, I want to break the loop. Wake, that's it. And now we need to flip the frame horizontally for the lty e, for the mirror review. So HemtiFrame equal to CV two dot flip method. Flip. Then inside the round dresses, I'm going to pass the frame and also going to pass the flip code. And I'm going to flip it horizontally, so I'm going to pass one. Then we need to convert this Bier frame into RGB color. For that, I'm going to declare another variable RGB under scope frame, equal to hemotype CV two dot CVD color. This one. Then inside the round dresses, first need to pass the source file, so I'm going to pass frame, then only to pass the colored code. CV two dot here I'm going to convert color, underscore BGR to RGB, this one. Here we convert this frame into, BGR to algeb. Now, after convert this frame into algeb color, we need to process the frame. For that, we need to run process method. Here I'm going to tie result equal to hands, Do process method. Process. Then inside this roundness, you need to pass the RGB frame. RGB, underscore free. Next, here I'm going to run I condition. If result got multi hand So type, multi underscoehand, hand, underscoe landmarks. Then insert this condition, I want to run a four loop. He type four, hand, uncoe landmarks in result, dot, multi hand landmarks. So I'm going to copy this line, multi hand landmarks. Another pesttive. Then inside this four loop, we need to draw the landmarks. Basically, I talk about these landmarks. We need to draw these landmarks. Here we check the detected hands in result variable. It contain the detected hand landmarks, which is written by media pipe hands module. If a hand is detected, this will hold a list of hand landmark objects. If no hands are detected, this value is none, and then we iterate through the detected hands using for for hand landmarks in result multi hand landmarks. As you know, each object contains 21 landmark points like thumb, TP, index finger tip, et cetera, and each of the landmark coordinate save XXs YX and ZXs value. And now we need to draw landmark in our hand. For that, here I'm going to type inside this follow, amino type, MP underscoe and drawing NP underscore drawing, dot, here, I'm going to draw the landmark. So I'm type, draw underscoe landmark. Draw landmark is that the rounder says, first, I'm going to provide the frame, in which frame I want to draw the landmark. So pass frame. Then we need to provide the hand landmarks, the hand landmark variable, and landmarks. This variable hand landmarks is basically the main object, which is going to contend 21 points. Next, we need to define which points are connected, like joins and tips. For that, I'm going to type MP underscore, Hence, draw the landmarks also into account the fingers. For that, I'm going to declare another variable fingered underscore count. Finger count equal to, I'm going to type a method, count finger. Is the rounder says, here I'm going to type hand landmark hand mark, this one. Basically, here I call the function, which I already create, this one, count finger. And as you can see as parameter, we need to pass hand landmarks. So we call this function. Now after count the finger, we need to display the finger count. For that, here, I'm going to use put text method. So type CV two dot, put text. Put text. Then inside the roundress first, I'm going to pass the frame with source image. Frame, comma, and here I'm going to use string method. Then inside the double coats, I'm going to type fingers, colon inside the Cibass, I'm going to pass this variable, Tinger count. This one. And next, we need to define the position of this text. So inside the round press, I'm going to pass 1070. Then you need to define the font style. So I'm going to tie CV to the font hoch simplex, underscoeh simplex, this one. Then you need to define the font size, so I'm going to pass two. It is the font scale value. Next, we need to define the font color. So as a font color, I'm going to pass blue color. We need to pass it in a BGR format. First, I'm going to type 2550, zero. After that, you need to pass the font thickness. As a phone thickness, I'm going to pass two. So we successfully put this text in this frame. After that, we need to display this frame. For that, we need to run, I'm ShomthodOside the IP condition. So here, outside the IP condition, I'm going to type CV two dot, CV two dot, I am Shoathod I am show. Then inside the rounds, inside the double course, I'm going to type, finger counted. Inside the double course, first, we need to define the windowame and our windowname is fingered counted. Then we need to pass the frame name, which frame I want to display our frame name is frame. Also we need to break the loop using weitey method. I type I, CV two dot wet key. Inside the round pieces, I'm going to wait for one millisecond and zero X F equal to equal to ORD inside the roundresses. Inside the single course, I'm going to pass Q. Then inside this IP condition, I want to break lute, break. Then I'm going to release the captured. Cap dot release. Also, we need to destroy all windows. Sum type, CV two dot, destroy all windows. This one. And before we run this code, let me show you something. Here, we need to make it draw underscore landmarks. We need to add a, also into type mphand dot hand connection. It is going to define which points are connected, like joints and tips, et cetera. After that, you can set this file and run this code in your terminal. And if you run this code in your terminal, at first, it's going to show some warning. This warning, then it's going to start your webcam. You need to wait for some time. As you can see, it start our web camera. Now I'm going to raise my closed hand. As you can see, now it show zero fingers and also it track all the fingers successfully. Here you can see all the landmarks and now I'm going to spread all of my finger. As you can see on your screen, now it print five finger. You can notice it track all the finger joints and teeth. And if I close my thumb, now you can see it print four finger. If I rise it again, again, it print five. And if I move my index finger, as you can see, now it print four finger. At the same way it track all the fingers. It count all the fingers and also it track all the finger coordinates. This is how our finger counter work, and if you want to close this program, then you need to praise small Q in your keyboard and it's going to break the loop. I hope you enjoy this project. Thanks for watching this video and stay tuned for our next project.