1 00:00:00,030 --> 00:00:06,230 Hey! Here we are again and in this lecture we will continue building our real-world 2 00:00:06,230 --> 00:00:14,599 motion detection program. Now before I go and write the code ,I would first like to 3 00:00:14,599 --> 00:00:20,880 explain to you the architecture of the program that we'll be building. So how 4 00:00:20,880 --> 00:00:25,920 this program is able to detect motion in the video. And I assume you already know 5 00:00:25,920 --> 00:00:32,640 how to build this script, so we built this in the previous lectures, and what 6 00:00:32,640 --> 00:00:44,250 the script does is that in case you've missed it. Hey! So what this does is it 7 00:00:44,250 --> 00:00:51,600 triggers the video as you saw from the computer webcam, so we were processing 8 00:00:51,600 --> 00:00:57,590 frames here in this while loop and so on. So you know this I will not go through 9 00:00:57,590 --> 00:01:04,229 them right now, so what we need to do next is we need to process those frames 10 00:01:04,229 --> 00:01:09,330 that are being iterated in this while loop, and I've got some pictures here to 11 00:01:09,330 --> 00:01:15,330 illustrate my ideas, my concepts. Let me go to the directory where they are located. 12 00:01:15,330 --> 00:01:22,110 So what this motion detection program will do is it will trigger the 13 00:01:22,110 --> 00:01:26,970 webcam just like our current script does and one condition for the program to 14 00:01:26,970 --> 00:01:31,740 work well is that once you trigger the webcam the first frame of the video 15 00:01:31,740 --> 00:01:37,290 should be a static background. So if you're planning to use this 16 00:01:37,290 --> 00:01:44,549 program let's say you'll set up a webcam on a laptop or as Raspberry Pi server. Let's 17 00:01:44,549 --> 00:01:50,869 say you want to detect the movement of of a certain animal in an area, so first 18 00:01:50,869 --> 00:01:56,189 you'd want to capture, to trigger the camera while the background is 19 00:01:56,189 --> 00:02:01,439 static and then you want to use this background as a base image so that you 20 00:02:01,439 --> 00:02:06,119 can compare the other images and then Python can detect if there is a change 21 00:02:06,119 --> 00:02:08,729 between the first frame and the next frames. 22 00:02:08,729 --> 00:02:14,810 So that's one thing you need to do and so this is an example of background. 23 00:02:14,810 --> 00:02:22,290 And then something, the animal will appear on your camera. Okay, and then what 24 00:02:22,290 --> 00:02:28,190 you have to write in your script in the program is that first we would want to 25 00:02:28,190 --> 00:02:33,989 gray out this image, so the background image at the current frame of the camera. 26 00:02:33,989 --> 00:02:38,940 So you'll store the first frame of the video capture in a variable, and then you 27 00:02:38,940 --> 00:02:44,940 will convert that frame to a grayscale image and then the while loop will 28 00:02:44,940 --> 00:02:49,709 go through the current frames and you'll do the same for the current frames, so you'll 29 00:02:49,709 --> 00:02:55,440 convert them to grayscale, and then what you do to these two grayscale images of 30 00:02:55,440 --> 00:03:00,660 the current iteration of the loop, you'll apply the difference between them. 31 00:03:00,660 --> 00:03:06,000 So this is a difference an example of a difference frame, of a delta frame if you 32 00:03:06,000 --> 00:03:10,680 can say like that. In this particular image you'll notice that behind me, there 33 00:03:10,680 --> 00:03:16,319 is a lamp of the room and normally you wouldn't be able to see the lamp, but Python 34 00:03:16,319 --> 00:03:21,090 is actually making the difference between the frame where I was appearing 35 00:03:21,090 --> 00:03:27,030 and the background frame where the lamp is visible, so it comes up with this gray 36 00:03:27,030 --> 00:03:32,150 image where each pixel has a certain value, so it has some intensity values. 37 00:03:32,150 --> 00:03:40,739 All right, so that means for instance the the high intensity values like this 38 00:03:40,739 --> 00:03:46,829 one where I am means there is potential motion in this area here, while 39 00:03:46,829 --> 00:03:52,819 the blacks areas imply there is no motion, but you also see some light black 40 00:03:52,819 --> 00:03:59,060 pixels here because when i when I appear on camera there is shadow behind me and so on. 41 00:03:59,060 --> 00:04:06,030 But we will apply something else later on which is called the threshold, so we'll 42 00:04:06,030 --> 00:04:09,450 basically say that if you see adifference in the delta frame, in the frame 43 00:04:09,450 --> 00:04:14,639 that I just showed you, if you see a difference of more than 100 intensity 44 00:04:14,639 --> 00:04:21,609 convert that pixels, those pixels to white pixels, okay. 45 00:04:21,609 --> 00:04:28,719 And for pixels that are below the threshold convert them to black, okay. 46 00:04:28,719 --> 00:04:34,300 So you come up with the outline of the object that is moving in the camera, 47 00:04:34,300 --> 00:04:40,210 in the frame. So we're doing all these processes, inside the while loop and then 48 00:04:40,210 --> 00:04:44,409 once we have calculated the threshold frame inside the loop what we'll do to the 49 00:04:44,409 --> 00:04:51,490 current frame is will find the contours of the white objects in the frame, okay. 50 00:04:51,490 --> 00:04:57,610 So for this particular image we would have contours around this object here, and the 51 00:04:57,610 --> 00:05:02,199 contours around this, and around this is as well, and then we'll write a for loop that will 52 00:05:02,199 --> 00:05:07,150 iterate through all the contours of the current frame, okay, it will go to this contur, 53 00:05:07,150 --> 00:05:11,919 to this, to this, and to this one, and then inside that loop what we will do, is we will 54 00:05:11,919 --> 00:05:18,279 check if the area of the contours, so this for example has an area of let's say 500 55 00:05:18,279 --> 00:05:24,430 pixels, so if the area of the contour is more than 500 pixels for 56 00:05:24,430 --> 00:05:30,159 example, then consider this as a moving moving object. If there is let's say less 57 00:05:30,159 --> 00:05:35,379 than 500 like this one here is probably 20 pixels let's say, this will not be 58 00:05:35,379 --> 00:05:41,169 considered a moving object, okay, I hope that makes sense, and then what we'll do 59 00:05:41,169 --> 00:05:46,900 next is we'll draw a rectangle around the contours that were greater than the 60 00:05:46,900 --> 00:05:53,020 minimum area and I will show those rectangles in original image, so in the 61 00:05:53,020 --> 00:05:58,389 color version of the current frame, and that means we will see a rectangle in the 62 00:05:58,389 --> 00:06:01,509 video while the video is playing we will see rectangle around the object. 63 00:06:01,509 --> 00:06:06,339 Later on we will detect the times that the object, the moving object entered 64 00:06:06,339 --> 00:06:11,830 the background, the video frame, and the time that the object exited the background. 65 00:06:11,830 --> 00:06:16,719 But for now let's simply focus on detecting the object, the moving object 66 00:06:16,719 --> 00:06:23,639 in the video. Okay. Let's go back to the script. 67 00:06:26,430 --> 00:06:34,770 Close this and this and this. Okay and let's remove some unnecessary lines here. 68 00:06:34,770 --> 00:06:39,180 So we had the a variable here which we created because we wanted to see how 69 00:06:39,180 --> 00:06:45,539 many frames we had in the video, so we don't need that anymore, and we don't 70 00:06:45,539 --> 00:06:51,000 need the script to stop for three seconds. So I remove the time.sleep. 71 00:06:51,000 --> 00:06:59,100 And this one as well. Okay, and now here the first part is the most trickiest one. 72 00:06:59,100 --> 00:07:04,650 We need to figure out a way to store the current frame of the video so as soon as 73 00:07:04,650 --> 00:07:10,380 the video starts we want to store that Numpy array in a variable and have that 74 00:07:10,380 --> 00:07:14,160 variable static, so we don't want to change the value of the variable while 75 00:07:14,160 --> 00:07:20,310 the while loop runs in the script. The way to do that is we would first 76 00:07:20,310 --> 00:07:28,259 need to create a variable for the frame, for the first frame, so we need to assign it 77 00:07:28,259 --> 00:07:35,340 a None value, so None is a special Python data type that allows you to create a 78 00:07:35,340 --> 00:07:39,270 a variable and assign nothing to it but you have the variable there, so if you call 79 00:07:39,270 --> 00:07:44,940 this variable later Python will not say variable is not defined. Okay, if you 80 00:07:44,940 --> 00:07:51,510 don't understand it, please hold on and you will get it in just a while. And what you 81 00:07:51,510 --> 00:07:57,650 need to do now is to write a conditional and apply a continue statement in there. 82 00:07:57,650 --> 00:08:02,580 Let me write the conditional first and then I'll explain to you what this does. 83 00:08:02,580 --> 00:08:17,880 So we want to check if the first frame is None and if it is None we will assign 84 00:08:17,880 --> 00:08:27,750 the first frame the gray frame. So what this does is that the scrip will 85 00:08:27,750 --> 00:08:33,870 run, the video will be triggered and then the while loop will start to run and it will get 86 00:08:33,870 --> 00:08:38,710 the first frame of the video, and it will store it in this frame variable, 87 00:08:38,710 --> 00:08:43,000 and this frame variable will be converted into a gray frame and then we 88 00:08:43,000 --> 00:08:48,520 say if the first frame is None, which is true in the first iteration of the loop 89 00:08:48,520 --> 00:08:53,140 this is true, so the first frame is actually None because we assigned None 90 00:08:53,140 --> 00:09:00,310 here, assign the gray numpy to the first frame. So the first frame will get 91 00:09:00,310 --> 00:09:06,310 the grayscale image which represents the very first frame of the video. So this 92 00:09:06,310 --> 00:09:12,730 happens in the very first iteration of the loop, okay? But then, what will happen is 93 00:09:12,730 --> 00:09:17,980 Python will execute these other lines of code and then it will go to the second loop 94 00:09:17,980 --> 00:09:23,470 and what Python will do is it will grab the second frame of the video, okay? 95 00:09:23,470 --> 00:09:28,810 Let's say the first frame was a background image, then suddenly an object appears in 96 00:09:28,810 --> 00:09:34,450 the camera, in front of the camera so Python will grab the numpy array that 97 00:09:34,450 --> 00:09:39,760 contains that object, so the second frame and this frame will be converted to gray 98 00:09:39,760 --> 00:09:46,060 and then here we say if first frame is None, the first frame variable will get the 99 00:09:46,060 --> 00:09:52,600 first frame of the video and once we have grabbed the first frame we don't 100 00:09:52,600 --> 00:09:57,700 want this other lines of code to be executed because here we will have, you 101 00:09:57,700 --> 00:10:01,900 know, we'll apply the difference between frames and we will blurry the frames and 102 00:10:01,900 --> 00:10:07,330 so on, so we don't want these to be executed, instead we want Python to go to 103 00:10:07,330 --> 00:10:12,760 the beginning of the loop and continue with the second frame. To do that we need 104 00:10:12,760 --> 00:10:19,779 to go and write continue here. So this means continue to the beginning of the loop 105 00:10:19,779 --> 00:10:26,829 and don't go and around the rest of the code, okay, so the first frame is None. 106 00:10:26,829 --> 00:10:31,900 The first frame gets this value of the first image of the video and then goes to the 107 00:10:31,900 --> 00:10:37,209 next iteration, and then the next iteration what will do, it will grab 108 00:10:37,209 --> 00:10:42,459 the second frame of the video and then it will calculate the gray version of that 109 00:10:42,459 --> 00:10:48,459 frame and then it goes again to the conditional, and in this case is the 110 00:10:48,459 --> 00:10:51,880 first frame None? No, it's not because the first frame got the value of the 111 00:10:51,880 --> 00:10:54,770 gray image in the first iteration of the 112 00:10:54,770 --> 00:11:00,380 while loop, so this lines here will not be executed at the second iteration of 113 00:11:00,380 --> 00:11:08,440 the loop, okay? Great. That means we can now apply Delta frame, 114 00:11:08,440 --> 00:11:13,580 so we can calculate the difference between the first frame and the current 115 00:11:13,580 --> 00:11:17,090 frame of the image. So the first frame is the first frame variable and the 116 00:11:17,090 --> 00:11:22,850 current frame is a gray variable, but before that we would like to do 117 00:11:22,850 --> 00:11:28,880 something to the current frame of the image. We want to apply a Gaussian blur 118 00:11:28,880 --> 00:11:38,210 to the image. And a dot here. So the reason we want to apply Gaussian 119 00:11:38,210 --> 00:11:42,950 blur is that we want to blur the image so we want to make it blurry so to 120 00:11:42,950 --> 00:11:48,260 smooth it because that removes noise and increases accuracy in the calculation 121 00:11:48,260 --> 00:11:54,110 of the difference, so this gets parameter the image you want to blur. 122 00:11:54,110 --> 00:11:59,030 So we are passing the gray image here and we're storing the blurry version of 123 00:11:59,030 --> 00:12:05,270 the image in a gray image again, and then we have another parameter which comes as 124 00:12:05,270 --> 00:12:12,710 a tuple and here we need to pass a width and height of a Gaussian kernel, so which 125 00:12:12,710 --> 00:12:18,560 is basically the parameters of the blurriness, but 21 would be accepted numbers. 126 00:12:18,560 --> 00:12:23,920 And you also need a number, last parameter so that would be the standard deviation 127 00:12:23,920 --> 00:12:31,550 and pass zero. Zero here is also commonly used. If you want to learn about them you 128 00:12:31,550 --> 00:12:36,830 can go through the documentation but these values would be good. So we're 129 00:12:36,830 --> 00:12:45,650 making the gray image blurry here, then down here now we need to compare the 130 00:12:45,650 --> 00:12:51,500 first frame of the image, so the background with the current frame. 131 00:12:51,500 --> 00:12:59,780 Let's call this delta frame and that would be equal to cv2.absdifference so 132 00:12:59,780 --> 00:13:04,430 absolute difference between the first frame 133 00:13:04,430 --> 00:13:10,010 and the current frame which is gray. Note that the first frame will also be 134 00:13:10,010 --> 00:13:15,740 a gray version, a blurry gray version actually, so we are comparing 135 00:13:15,740 --> 00:13:23,930 here two blurriede grayscale images, okay? And what this will give us is another 136 00:13:23,930 --> 00:13:29,920 image, okay? And actually I would like to show that image here on the screen. 137 00:13:29,920 --> 00:13:42,530 Cv2image show Delta frame. Let's see what we'll get out of this. And before running 138 00:13:42,530 --> 00:13:47,480 the script I'll disappear from the view first and then will appear again. 139 00:13:47,480 --> 00:13:55,720 Let's see. Okay, nothing happened because. I forgot to enter the name of the window 140 00:13:55,720 --> 00:14:07,160 Let's call this delta frame and let's say gray frame for this. Now let me run 141 00:14:07,160 --> 00:14:22,520 it again. And here so this is the blurred grayscale version, this one here and we 142 00:14:22,520 --> 00:14:33,010 have the difference, okay, so you can see the lamp behind me in the delta frame. Great. 143 00:14:33,010 --> 00:14:42,380 Press the Q key and quit the video now if you if I want to print just to check 144 00:14:42,380 --> 00:14:52,220 the delta frame, this will allow us to see the difference between intensities 145 00:14:52,220 --> 00:15:06,529 of the corresponding pixels so let's see. And if I quit this now 146 00:15:06,529 --> 00:15:15,029 here we go. So what we have here five means there is no difference, so in this 147 00:15:15,029 --> 00:15:20,970 area there's no motion probably but then you have 174 which is quite white, 148 00:15:20,970 --> 00:15:28,769 so that means Python will classify this as motion. Okay, that was just to show 149 00:15:28,769 --> 00:15:37,529 you the values of the delta frame. What we need now is to classify these values. 150 00:15:37,529 --> 00:15:41,940 So let's say we want to assign a threshold, so let's say if you have 151 00:15:41,940 --> 00:15:45,899 values that are more than 30, so if the difference between the first frame 152 00:15:45,899 --> 00:15:52,860 and the current frame is more than 30, we will classify that as white. 153 00:15:52,860 --> 00:15:58,019 So we will say there's probably motioning in those pixels, so there's an object in 154 00:15:58,019 --> 00:16:03,779 those pixels, and if the difference is less than thirty we'll assign it black 155 00:16:03,779 --> 00:16:12,180 pixels, okay? So we can do that so we are here, we can do that using the 156 00:16:12,180 --> 00:16:20,000 threshold method of the cv2 library. Let's say thresh_delta equals to 157 00:16:20,000 --> 00:16:31,019 cv2.threshold, and what this expects is a image that you want to threshold and then you 158 00:16:31,019 --> 00:16:36,839 want to specify a threshold limit. So 30, we said 30. And what color do 159 00:16:36,839 --> 00:16:40,860 you want to assign to the values that are more than 30? Well we want to saw 160 00:16:40,860 --> 00:16:48,570 in a white color which corresponds to the 255 value, okay? And you also need 161 00:16:48,570 --> 00:16:56,490 one more argument here which is the threshold method. There are quite a few 162 00:16:56,490 --> 00:17:03,209 methods out there, but we are using this method here, so threshold binary. 163 00:17:03,209 --> 00:17:06,949 You can experiment with others if you like. Great and now let's see how this thresh 164 00:17:10,860 --> 00:17:17,189 Delta frame will look like. Cv2.imshow threshold frame. 165 00:17:17,189 --> 00:17:28,769 Okay, thresh_delta. Great, let's see! 166 00:17:31,620 --> 00:17:42,909 apparently I wrote this wrongly. Delta frame, not frame Delta. And one more 167 00:17:42,909 --> 00:17:52,450 typo in line 19 which is here, cv2, okay. Let's hope this time will work, and yet 168 00:17:52,450 --> 00:18:01,210 another error. And this time I've missed something here. Okay, this method here so 169 00:18:01,210 --> 00:18:06,610 the threshold method actually returns a tuple with two values, and the first 170 00:18:06,610 --> 00:18:11,950 value is it's needed when you use other threshold methods, so the first item or 171 00:18:11,950 --> 00:18:16,179 the topple basically suggests a value for threshold if you're using other 172 00:18:16,179 --> 00:18:21,100 methods, but for financial binary you only need to access the second item of 173 00:18:21,100 --> 00:18:27,490 the tuple which is the actual frame that is returned from the threshold 174 00:18:27,490 --> 00:18:33,820 method, so you want to access the second item of the tuple. Okay, I promise this 175 00:18:33,820 --> 00:18:49,419 is a last error. Great, finally! And this is the threshold frame, so you can see my 176 00:18:49,419 --> 00:18:54,789 outline there, but you also see some shadows. So I am being detected as an 177 00:18:59,740 --> 00:19:05,409 object, but my shadows as well are being detected as on objects too. Okay. 178 00:19:05,409 --> 00:19:11,019 So you get the idea. Now we can go right away and use this Thresh Delta frame 179 00:19:11,019 --> 00:19:16,629 which I called it thresh Delta. I should call it the thresh frame, you know, just 180 00:19:16,629 --> 00:19:23,889 for name consistency. So we can go ahead now and create contours of the white 181 00:19:23,889 --> 00:19:29,590 objects in the thresh frame, but before that I would like to do something else. 182 00:19:29,590 --> 00:19:33,670 I would like to dilate those areas. So I want to remove the black holes 183 00:19:33,670 --> 00:19:39,490 from those big white areas in the image, so basically I want to smooth my 184 00:19:39,490 --> 00:19:45,550 threshold frame, and to do that you need to use the dilate method of the cv2 185 00:19:45,550 --> 00:19:56,680 library, so let's say we want to change the threshold frame. Cv2.dilate. Again you 186 00:19:56,680 --> 00:20:02,350 want to pass the threshold frame there. Now if you have a kernel array and you want 187 00:20:02,350 --> 00:20:07,270 this process to be very sophisticated you'd pass that array in here. We don't 188 00:20:07,270 --> 00:20:12,550 have any and we don't need one, so you need to pass None for this parameter and 189 00:20:12,550 --> 00:20:18,430 there's yet another parameter here. Iterations and let's say two, so this 190 00:20:18,430 --> 00:20:23,620 defines how many times do you want to go through the image to remove those holes. 191 00:20:23,620 --> 00:20:29,110 So the bigger this number is, the smoother this image will be. Okay. Now let me 192 00:20:29,110 --> 00:20:37,060 check it quickly. Oh yeah, I changed the name earlier, but I didn't change it 193 00:20:37,060 --> 00:20:54,750 here in the imshow method. Thresh frame. Okay, so you can probably notice that 194 00:20:54,750 --> 00:21:04,600 the areas, the white areas are smoother now so if I go away you'll only notice a few areas 195 00:21:04,600 --> 00:21:11,200 that are there because of my shadow, so it seems to be working so far. Let's quit 196 00:21:11,200 --> 00:21:19,750 it now. Great, so we've got these three frames and what's next? Well next is we 197 00:21:19,750 --> 00:21:26,920 need to find the contours of this dilated threshold frame. Regarding 198 00:21:26,920 --> 00:21:33,700 contour detection with openCV you have two methods, so you have a find contours 199 00:21:33,700 --> 00:21:40,240 and a draw contours method. And with the find conturs method what you do is you 200 00:21:40,240 --> 00:21:45,470 find the contours in your image and you store them in a tuple, on the other hand 201 00:21:45,470 --> 00:21:51,590 the draw contours method draws contours in an image, so in this case what we want to do 202 00:21:51,590 --> 00:21:57,920 is we want to find the contours and then we want to check if the area of this 203 00:21:57,920 --> 00:22:02,960 contour, so let's say you have a contour like a circle and so you want to find 204 00:22:02,960 --> 00:22:08,780 the area that this also defines, so you want to store those contours in 205 00:22:08,780 --> 00:22:13,910 a tuple and that's quite a weird syntax here, but bear with me. 206 00:22:13,910 --> 00:22:20,660 So you want to write underscore comma cnts comma and another underscore. 207 00:22:20,660 --> 00:22:27,200 And if you were in OpenCV2 you wouldn't need the first underscore and 208 00:22:27,200 --> 00:22:32,960 the first comma, so you'd have this in OpenCV2 with Python 2. If you 209 00:22:32,960 --> 00:22:39,050 have OpenCv3 with Python 3 this is what you need to write. Okay, and that 210 00:22:39,050 --> 00:22:48,980 would be equal to cv2.findContours. And then you want to pass afraid that 211 00:22:48,980 --> 00:22:55,940 you want to find the contours for. And it's good to actually use the copy of the frame 212 00:22:55,940 --> 00:23:02,480 so you don't want to modify the threshold frame, so use copy here. 213 00:23:02,480 --> 00:23:09,020 So this is the first parameter, the frame you want to find contours from, and then 214 00:23:09,020 --> 00:23:18,950 you have the method retrieve external. So you want to draw the external contours 215 00:23:18,950 --> 00:23:24,080 of the objects that you'll be fining in the image, and you've got yet another 216 00:23:24,080 --> 00:23:34,700 argument. Chain_aprox_simple. So this is approximation method that OpenCV will 217 00:23:34,700 --> 00:23:42,890 apply for retriving the conturs, great. So what we have is we are iterating 218 00:23:42,890 --> 00:23:47,240 through the current frame so we are blurring it, and converting it to 219 00:23:47,240 --> 00:23:53,650 grayscale, and find a data frame, and apply the thresholds, so the black 220 00:23:53,650 --> 00:23:59,830 black and white image, and then we find all the contours of the objects, of the 221 00:23:59,830 --> 00:24:05,860 distinct objects in this image so if you've got two white continuous areas in 222 00:24:05,860 --> 00:24:10,660 your image but they are distinct, you'll get two contours, so one contour for each of 223 00:24:10,660 --> 00:24:18,010 the areas, and these corners will be stored in this, cnts variable. So we're talking 224 00:24:18,010 --> 00:24:22,510 about the current frame, and that's what we want to do, is we want to filter out 225 00:24:22,510 --> 00:24:27,310 these contours, so we want to check that we want to keep only the contours that are, 226 00:24:27,310 --> 00:24:33,550 let's say that have an area that is bigger than 100, 1000 pixels. For that you 227 00:24:33,550 --> 00:24:50,520 need to iterate. Let's say for contour in cnts, and if cv2.contour area of 228 00:24:50,520 --> 00:24:57,280 the contour, so the contour that we are iterating through, so if this is less 229 00:24:57,280 --> 00:25:06,330 than 1000, continue to the beginning of the for loop again. So what this means is 230 00:25:06,330 --> 00:25:11,530 let's say Python found three contours and it will go through the first one and it 231 00:25:11,530 --> 00:25:15,700 will say if the area of this contour, so we use the contour area of the cv2 232 00:25:15,700 --> 00:25:22,870 library, if the area has less than 1000 pixels, go to the next contour, so go 233 00:25:22,870 --> 00:25:28,240 to the second contour and check again, and again, and again. Otherwise if the area 234 00:25:28,240 --> 00:25:33,700 is bigger than, or equal to 1000, the next lines here after the for loop will be 235 00:25:33,700 --> 00:25:41,460 executed. So what do you want to do if a contour is greater than 1000 pixels? 236 00:25:41,460 --> 00:25:47,290 Well we want to draw the rectangle surrounding that contour to the 237 00:25:47,290 --> 00:25:54,970 current frame, so make sure you are inside the for loop, 238 00:26:00,180 --> 00:26:08,070 So these are the parameters that define the rectangle and that would be equal to 239 00:26:08,070 --> 00:26:19,740 cv2.boundRectangle of the current contour, so if the contour is equal or 240 00:26:19,740 --> 00:26:25,110 greater than 1000 pixels, so if I it has an a of equal, or greater 241 00:26:25,110 --> 00:26:31,730 than 1000 pixels, this will be executed. So we are creating a rectangle and then 242 00:26:31,730 --> 00:26:37,910 we want to draw that rectangle to our frame, to our current frame so cv2 243 00:26:37,910 --> 00:26:44,780 rectangle, so we already consumed this method in our face detection lectures. 244 00:26:44,780 --> 00:26:51,720 And here we would want to pass the color frame. Okay, so that is the frame 245 00:26:51,720 --> 00:26:59,490 and we want to specify x and y here, so these are the coordinates of upper left corner of 246 00:26:59,490 --> 00:27:06,420 the rectangle. X, y and you want to specify the coordinates of the right 247 00:27:06,420 --> 00:27:18,480 lower corner of the rectangle as well so X W, just like that. And also the color of 248 00:27:18,480 --> 00:27:27,720 your rectangle. Let's say green and the width. Let's say 3, so what we did in these two 249 00:27:27,720 --> 00:27:35,520 lines is that we created this tuple with this four coordinates and these values, 250 00:27:35,520 --> 00:27:41,520 these will be assigned automatically, so x and y will get the value from 251 00:27:41,520 --> 00:27:47,490 the rectangle bounding this contour, this current contour of the for loop. 252 00:27:47,490 --> 00:27:51,060 And then these values will be used to draw a rectangle in the frame, in the current 253 00:27:51,060 --> 00:27:56,880 frame, and then we want to show that current frame. So let me edit here. 254 00:27:56,880 --> 00:28:04,850 Image show. Let's call this color frame. And frame. 255 00:28:04,850 --> 00:28:13,700 Actually this method here is boundingRect. Let me try the script now. 256 00:28:16,810 --> 00:28:26,660 And cv2 has no attribute find countours. co here I've got a u there 257 00:28:26,660 --> 00:28:38,300 that shouldn't be there. Let's try it again. Yep, that's funny. 258 00:28:38,300 --> 00:28:51,230 I tend to mistype this word. Countour area in line 24 and remover the u. It should be 259 00:28:51,230 --> 00:28:55,120 contour here as well. okay, countour is not defined again. I've 260 00:29:06,440 --> 00:29:19,130 got another one here, so bear with me. Try again and yeah, this time seems to be 261 00:29:19,130 --> 00:29:33,260 working. So no objects, objects. No objects, objects, great. Great, so that's what I 262 00:29:33,260 --> 00:29:37,760 want to teach you in this lecture and we'll continue in the next lectures 263 00:29:37,760 --> 00:29:42,500 because what we've done so far is that we can detect that object and we 264 00:29:42,500 --> 00:29:47,030 can draw a rectangle around that object, but this is not very practical. I mean in 265 00:29:47,030 --> 00:29:52,010 real world it's not enough to just draw a rectangle around your object 266 00:29:52,010 --> 00:29:55,430 and then that's it. So what we'll be doing in the next 267 00:29:55,430 --> 00:30:01,040 lecture is will be storing the times that the object enters the frame and when 268 00:30:01,040 --> 00:30:07,130 the object exits the frame, so we've got some more lines to add to this code and I 269 00:30:07,130 --> 00:30:13,880 know this code here was quite a lot to consume, but I hope these are clear now. 270 00:30:13,880 --> 00:30:18,830 If you have any questions please feel free to ask them and I'll see you 271 00:30:18,830 --> 00:30:19,000 in the next lecture.