1 00:00:02,640 --> 00:00:09,810 In this video tutorial, we will learn how we can create each of the detected object using Deepsort 2 00:00:09,840 --> 00:00:11,400 object tracking algorithm. 3 00:00:11,970 --> 00:00:18,990 We will be integrating Yolo v9 with Deepsort object tracking algorithm, so we will use YOLO v9 to detect 4 00:00:18,990 --> 00:00:22,410 objects in image video or in a live webcam feed. 5 00:00:22,860 --> 00:00:30,060 And then we will be using Deepsort object tracking algorithm to track each of the detected object until 6 00:00:30,060 --> 00:00:34,890 that object remains in the frame of a video or in a live webcam feed. 7 00:00:35,280 --> 00:00:37,200 So what we do in object tracking? 8 00:00:37,500 --> 00:00:42,480 In object tracking, we basically assign a unique ID to each of the detected object. 9 00:00:42,480 --> 00:00:48,600 And then using that unique ID, we track that object until that object remains enough. 10 00:00:49,170 --> 00:00:54,450 Uh, in remains in, uh, in the frame of a video or in a live webcam feed. 11 00:00:54,690 --> 00:00:59,850 So using that unique ID that we assigned to each of the detected object, we then track the object. 12 00:00:59,850 --> 00:01:05,640 We can also draw trails, follow the path that check the path the object is following. 13 00:01:05,820 --> 00:01:07,890 So that's all we are going to learn. 14 00:01:07,890 --> 00:01:12,420 In this tutorial we will be integrating YOLO v9 with Deepsort object tracking algorithm. 15 00:01:12,420 --> 00:01:19,110 You all know your V9 was released last month on 21st February 2024, which is the state of the art object 16 00:01:19,320 --> 00:01:25,200 detection algorithm and YOLO be nine outperforms all the previous object detection algorithms as well. 17 00:01:25,200 --> 00:01:29,910 While Deepsort algorithm is the state of the art object tracking algorithm. 18 00:01:29,910 --> 00:01:35,070 So by combining the state of the art object track detection algorithm with object tracking algorithm, 19 00:01:35,070 --> 00:01:38,610 we will be doing object detection last object tracking. 20 00:01:38,610 --> 00:01:40,860 So that's what we are going to learn. 21 00:01:41,720 --> 00:01:42,860 In this tutorial. 22 00:01:42,860 --> 00:01:46,670 Thus, we will also see if the object gets blurred or hidden. 23 00:01:46,700 --> 00:01:51,620 We will also be tracking that object and that object unique ID will remain the same. 24 00:01:52,100 --> 00:01:54,200 Okay, so let's get started. 25 00:01:54,200 --> 00:02:00,560 So over here you can see in my jump I have opened PyCharm over here and I have Python 3.10 installed. 26 00:02:00,560 --> 00:02:02,150 You can see over here. 27 00:02:02,150 --> 00:02:07,100 Plus I have just created a project over here by the name of init Deepsort. 28 00:02:07,520 --> 00:02:10,100 So I have created this project in my app directory. 29 00:02:10,100 --> 00:02:14,510 So in the same way you can create this any project in any of your directory. 30 00:02:14,510 --> 00:02:16,460 Plus you can also use Visual Studio. 31 00:02:16,460 --> 00:02:17,690 I'm using PyCharm. 32 00:02:17,690 --> 00:02:20,120 You can also use Visual Studio as well. 33 00:02:21,590 --> 00:02:23,930 Between the first step, we'll go step by step. 34 00:02:23,930 --> 00:02:29,660 So if you just see this note while I've written all the steps, so which I will be pointing. 35 00:02:29,660 --> 00:02:32,120 So first we need to clone the repo. 36 00:02:32,690 --> 00:02:34,790 So I will just open the browser over here. 37 00:02:35,450 --> 00:02:35,630 It. 38 00:02:37,820 --> 00:02:40,520 And I will just go over here and. 39 00:02:42,010 --> 00:02:45,100 I will just write your lovin line GitHub over here. 40 00:02:46,410 --> 00:02:46,710 Lingua. 41 00:02:46,710 --> 00:02:47,520 Volume nine. 42 00:02:47,550 --> 00:02:48,270 Get up. 43 00:02:53,890 --> 00:02:57,340 So you'll just search for this guadalmina and get up over here. 44 00:02:57,340 --> 00:03:00,310 And I will just open this post link from here. 45 00:03:01,240 --> 00:03:04,300 Okay, so this is the Yolo v9 official GitHub repo. 46 00:03:04,330 --> 00:03:06,100 It has around 7000 stars. 47 00:03:06,100 --> 00:03:12,370 So I will just, uh, clone this complete, uh, repo into my, uh, PyCharm directory. 48 00:03:12,550 --> 00:03:14,830 So I will just click copy this link from here. 49 00:03:14,830 --> 00:03:17,470 Then I will just go to the PyCharm over here. 50 00:03:17,710 --> 00:03:19,870 And then I will just here. 51 00:03:20,500 --> 00:03:22,330 And then I will just write git clone. 52 00:03:22,330 --> 00:03:24,580 And here I will just add the repo link. 53 00:03:24,580 --> 00:03:29,140 So now it will clone this complete repo into my local directory over here. 54 00:03:29,140 --> 00:03:30,790 And you will find this okay. 55 00:03:30,790 --> 00:03:31,750 So it's done now. 56 00:03:32,320 --> 00:03:36,820 So if I just refresh this so you can find this complete Golovnin repo over here. 57 00:03:38,440 --> 00:03:43,660 So now in the step number two I will just start this uh yolo benign repo, uh, which I have cloned 58 00:03:43,660 --> 00:03:44,950 as my current directory. 59 00:03:44,950 --> 00:03:48,430 So I will write CD yolo VI over here okay. 60 00:03:50,920 --> 00:03:51,310 Bernard. 61 00:03:51,310 --> 00:03:52,510 This is being. 62 00:03:55,060 --> 00:04:01,360 So now if I just open the requirements.txt file, the requirements.txt file you can find, I have the, 63 00:04:01,360 --> 00:04:07,390 they have listed all the required packages that are required uh, to run the repo successfully. 64 00:04:07,600 --> 00:04:08,140 Okay. 65 00:04:08,140 --> 00:04:10,240 So okay. 66 00:04:10,240 --> 00:04:14,350 So I will just uh, as we will be integrating object tracking using Deepsort algorithm. 67 00:04:14,350 --> 00:04:17,980 So we need to add few more uh, packages as well. 68 00:04:17,980 --> 00:04:24,220 So which is you can add it and plus I will see if the numpy package is over here. 69 00:04:24,280 --> 00:04:28,600 So okay so I will require numpy package. 70 00:04:31,660 --> 00:04:32,560 Less than. 71 00:04:34,080 --> 00:04:35,460 1.24. 72 00:04:35,730 --> 00:04:37,380 And so we will be getting an error. 73 00:04:37,380 --> 00:04:38,490 So it's always factored. 74 00:04:39,180 --> 00:04:47,790 So this and we also required g down package because we will be uh installing demo videos or deepsort 75 00:04:47,790 --> 00:04:50,640 files from the drive into the tracker into this PyCharm. 76 00:04:50,850 --> 00:04:52,470 Uh ID okay. 77 00:04:52,470 --> 00:04:58,350 So now in the step number two, what is install all the required packages okay. 78 00:04:58,350 --> 00:05:01,530 So we will be installing all the required packages over here. 79 00:05:06,250 --> 00:05:09,730 Well, now to install all these required packages I will simply write over here. 80 00:05:09,760 --> 00:05:15,370 Pip install minus r requirements dot txt. 81 00:05:15,370 --> 00:05:20,740 So we don't need to install all these packages one by one if we just write requirements dot txt. 82 00:05:20,770 --> 00:05:21,550 Pip install. 83 00:05:21,640 --> 00:05:26,260 It will install all the packages that are required listed in this requirements.txt file. 84 00:05:26,260 --> 00:05:28,030 And I will now click on enter. 85 00:05:28,540 --> 00:05:32,440 So now all these packages will be get installed over here. 86 00:05:40,820 --> 00:05:44,330 So this will take some time, but all these packages to get installed. 87 00:05:44,720 --> 00:05:49,640 So let's wait for these packages to get installed and then we will go ahead. 88 00:06:01,270 --> 00:06:03,790 Now you can see that the packages are installed. 89 00:06:03,790 --> 00:06:08,620 So now in the step number three, what we need to do is we need to download a Deepsort files. 90 00:06:08,620 --> 00:06:12,160 So I will just copy this link from here. 91 00:06:13,700 --> 00:06:14,090 Yes. 92 00:06:14,090 --> 00:06:14,810 Copy this. 93 00:06:16,990 --> 00:06:17,500 Okay. 94 00:06:17,500 --> 00:06:23,380 And I will just paste this link over here and I just click on enter. 95 00:06:23,380 --> 00:06:29,380 So this will download the Deepsort files from the drive into this Google Colab notebook or into this 96 00:06:29,380 --> 00:06:30,190 PyCharm ID. 97 00:06:30,370 --> 00:06:36,250 So I've already placed this deepsort zip file into my drive, and I'm directly downloading the Deepsort 98 00:06:36,250 --> 00:06:40,450 PyTorch zip file, uh, from my drive into this PyCharm IDE. 99 00:06:40,720 --> 00:06:44,050 Uh, so now we just need to unzip this deepsort files. 100 00:06:44,050 --> 00:06:51,520 So what I will do for this, I will just, uh, click over here and create a new script.py file. 101 00:06:52,540 --> 00:06:54,100 Uh, in the script or by file. 102 00:06:54,100 --> 00:06:58,810 I will just write some code, uh, to zip, uh, unzip these deepsort files. 103 00:07:05,340 --> 00:07:06,900 So over here. 104 00:07:10,170 --> 00:07:10,980 Import. 105 00:07:10,980 --> 00:07:13,710 I will just write over here import zip file. 106 00:07:16,030 --> 00:07:17,710 With zip file. 107 00:07:20,140 --> 00:07:20,410 Uh. 108 00:07:22,360 --> 00:07:23,410 Not zip file. 109 00:07:27,450 --> 00:07:32,130 Here I will just write the name which is deepsort by dot dot zip. 110 00:07:34,080 --> 00:07:35,700 But we need to read this. 111 00:07:37,590 --> 00:07:41,070 As the dash ref. 112 00:07:43,750 --> 00:07:46,900 And we need to extract all these complete files. 113 00:07:46,900 --> 00:07:48,370 So just do this. 114 00:07:48,610 --> 00:07:53,170 And over here I will just run this script file. 115 00:07:55,160 --> 00:07:57,050 And let's see how does it works. 116 00:07:57,080 --> 00:07:57,650 Okay. 117 00:07:57,650 --> 00:07:59,180 So do we have the okay. 118 00:07:59,180 --> 00:08:03,800 So now you can see I've just, uh, completely unzipped this file over here. 119 00:08:04,040 --> 00:08:04,640 Okay. 120 00:08:04,640 --> 00:08:07,670 So there is a function in Python unzip. 121 00:08:07,670 --> 00:08:10,580 But that function does not work in PyCharm. 122 00:08:10,580 --> 00:08:13,040 I have tried that function in Google Colab. 123 00:08:13,040 --> 00:08:15,770 It works fine there, but that doesn't work in PyCharm. 124 00:08:15,770 --> 00:08:19,880 So in the deepsort PyTorch, uh, if I have not unzipped it. 125 00:08:19,880 --> 00:08:25,550 So you can see here we have a Deepsort folder where we have the uh, complete deepsort.py file code 126 00:08:25,550 --> 00:08:26,120 as well. 127 00:08:26,120 --> 00:08:29,390 And we have the configuration and readme file over here as well. 128 00:08:29,390 --> 00:08:32,660 So this is what we all require okay. 129 00:08:33,530 --> 00:08:39,260 Next what we need to do is we need to download a demo video from here. 130 00:08:39,260 --> 00:08:41,660 So I've already added the link over here. 131 00:08:41,660 --> 00:08:46,430 So okay so then we are just uh good as well. 132 00:08:46,430 --> 00:08:50,720 So now in the terminal I will just write add the link over here. 133 00:08:50,720 --> 00:08:56,150 So this will download a demo video from a drive into this PyCharm IDE over here. 134 00:08:57,220 --> 00:08:57,550 Okay. 135 00:08:57,550 --> 00:09:02,470 So now you can see we have the demo video over here as well. 136 00:09:02,470 --> 00:09:06,310 So if I just re okay so no need to. 137 00:09:06,310 --> 00:09:08,500 So you can see here test dot mp4. 138 00:09:08,500 --> 00:09:13,990 Uh we have this demo video over here as well on which we will be doing detection and tracking okay. 139 00:09:15,810 --> 00:09:23,910 So in next part I will do is um, so in the detect.py we have all the object detection code. 140 00:09:23,910 --> 00:09:30,390 So I will just add another file detect dash du that tracking.py in which we will be doing object detection 141 00:09:30,390 --> 00:09:32,040 plus tracking over here. 142 00:09:35,230 --> 00:09:39,670 Over here you can see that I have added detect-dual-tracking.py file. 143 00:09:39,670 --> 00:09:42,610 So here I will be uh, adding the deepsort code. 144 00:09:42,610 --> 00:09:48,310 I have added some deepsort code over here which include initialization initialization of Deepsort algorithm. 145 00:09:48,310 --> 00:09:54,730 So over here what I have done for now is I have initialized or I have imported all the Deepsort required 146 00:09:54,730 --> 00:09:57,430 packages over here, which you can see over here. 147 00:09:57,430 --> 00:10:00,190 And then I have initialized Deepsort over here. 148 00:10:00,190 --> 00:10:05,770 I have initialized the Deepsort tracker over here, and all the parameters which you can find over here. 149 00:10:05,770 --> 00:10:09,160 So the important parameter we should, which you should need to know. 150 00:10:09,160 --> 00:10:12,700 I have uh, written all the comments about each of the parameter above. 151 00:10:12,700 --> 00:10:14,050 You can read that as well. 152 00:10:14,050 --> 00:10:16,540 So here is the maximum age parameter. 153 00:10:16,540 --> 00:10:19,480 So you should uh need to know about that parameter. 154 00:10:19,480 --> 00:10:26,800 So in the maximum age uh what we define is that for example uh due to an occlusion or if uh, for any, 155 00:10:26,800 --> 00:10:32,770 in any case, if an object tracking ID is lost for an object, um, so in the maximum age we define 156 00:10:32,770 --> 00:10:36,730 how many frames it should wait to get that, uh, ID back. 157 00:10:36,730 --> 00:10:43,570 So, uh, for example, if there is a car and we have lost that, uh, car object tracking ID due to 158 00:10:43,570 --> 00:10:48,610 an illusion, so how many frames that it should wait for it to get back? 159 00:10:48,610 --> 00:10:53,650 Or if it doesn't get back in this much frames, then it will assign a new ID. 160 00:10:53,680 --> 00:10:57,940 So for example, if I set the maximum age value to 20. 161 00:10:57,940 --> 00:11:05,290 So it will wait 20 frames, uh, to get the object tracking ID back, and if in 20 frames it does not 162 00:11:05,290 --> 00:11:11,020 get the object tracking ID back for that specific object, it will assign that object a new ID. 163 00:11:11,020 --> 00:11:17,530 So if you can see over here if an object tracking ID is lost, this parameter determines how many frames 164 00:11:17,530 --> 00:11:18,850 the tracker should wait. 165 00:11:18,910 --> 00:11:24,790 Wait before assigning a new ID so you can see all the details over I have written over here. 166 00:11:26,210 --> 00:11:32,480 Then over here I have to define the minimum confidence, maximum distance and the RMS maximum overlap 167 00:11:32,480 --> 00:11:34,070 maximum IOU distance. 168 00:11:34,070 --> 00:11:35,960 So you can read all those comments. 169 00:11:36,320 --> 00:11:38,300 I define every parameter over here. 170 00:11:38,450 --> 00:11:41,270 This will take quite some time so I will skip this. 171 00:11:41,270 --> 00:11:44,420 And here I've just created a function initialized Deepsort. 172 00:11:44,420 --> 00:11:47,120 So I have just uh initialized that over here. 173 00:11:47,360 --> 00:11:47,900 Okay. 174 00:11:47,900 --> 00:11:53,090 And here I have uh, because our YOLO v nine model has been trained on the Coco data set. 175 00:11:53,090 --> 00:11:55,850 So the Coco dataset consists of 80 different classes. 176 00:11:55,850 --> 00:11:59,090 So I have written all those 80 different classes names over here. 177 00:11:59,090 --> 00:12:02,780 And then uh, for example here I've defined the color label. 178 00:12:02,780 --> 00:12:09,560 So if we have the if the if in a video frame or in a live webcam feed, uh, person is detected, then 179 00:12:09,560 --> 00:12:15,740 we have assigned, uh, if the person is detected in a live webcam feed or in a video, then the bounding 180 00:12:15,740 --> 00:12:17,840 boxes will have the following color. 181 00:12:17,840 --> 00:12:21,560 So if the car is detected, then the bounding boxes will have the following color. 182 00:12:21,560 --> 00:12:27,170 If the motorbike is detected in a video frame image or in a live webcam feed, then the bounding boxes 183 00:12:27,170 --> 00:12:28,370 will have the following color. 184 00:12:28,370 --> 00:12:33,590 And if the box bus is detected in a video frame or in a live webcam feed, then the bounding boxes will 185 00:12:33,590 --> 00:12:34,400 have this color. 186 00:12:35,970 --> 00:12:36,390 Okay. 187 00:12:36,480 --> 00:12:39,810 And over here, uh, they we have loaded the model. 188 00:12:39,810 --> 00:12:42,060 We have applied non-expression over here. 189 00:12:42,060 --> 00:12:44,370 So this is all what is going over here? 190 00:12:44,370 --> 00:12:48,450 So from now where I will just add the object tracking code. 191 00:12:48,450 --> 00:12:49,470 Let me show you. 192 00:12:49,470 --> 00:12:53,400 So from here I will start adding the object tracking code over here. 193 00:12:54,820 --> 00:12:59,230 But before I add this so you can see over here okay. 194 00:12:59,380 --> 00:13:04,210 So so in the YOLO v nine we get the output in this form. 195 00:13:04,210 --> 00:13:08,530 Like you can see that uh, the Yolo v nine object detection algorithm returns the output. 196 00:13:08,530 --> 00:13:10,720 In this form we have the bounding box coordinates. 197 00:13:10,720 --> 00:13:15,880 We have the uh top right corner uh top left corner bounding box coordinate. 198 00:13:15,880 --> 00:13:19,120 You can see this is the top left corner and this is the bottom right corner. 199 00:13:19,120 --> 00:13:26,260 So the in the yolo uh, the yolo v nine uh, model outputs return the top left corner bounding box coordinate 200 00:13:26,260 --> 00:13:29,950 and the bottom right corner bounding box coordinates. 201 00:13:29,950 --> 00:13:35,740 So this is what yolo v nine returns top left corner bounding box coordinates and the bottom bottom right 202 00:13:35,890 --> 00:13:37,690 corner bounding box coordinates. 203 00:13:37,690 --> 00:13:43,030 So we need to pass the Yolo v nine object detection algorithm output to the Deepsort algorithm. 204 00:13:43,030 --> 00:13:50,260 So the Deepsort object tracking algorithm, uh, accepts uh the input in the form of width height of 205 00:13:50,260 --> 00:13:51,310 the bounding box. 206 00:13:51,760 --> 00:13:54,160 So this is the width okay. 207 00:13:54,160 --> 00:13:54,820 Here I'm wrong. 208 00:13:54,820 --> 00:13:55,960 This is the height. 209 00:13:55,960 --> 00:13:59,620 So this is wrong if I just, uh, correct this up over here. 210 00:14:00,220 --> 00:14:00,610 Okay. 211 00:14:00,610 --> 00:14:02,290 So this is. 212 00:14:03,320 --> 00:14:05,780 Uh, so let me just correct this. 213 00:14:06,410 --> 00:14:08,360 So this is basically. 214 00:14:09,870 --> 00:14:15,270 The height, and this is the width of the bounding box. 215 00:14:15,270 --> 00:14:21,810 So the Deepsort object tracking algorithm accepts input in the form of height, width of the bounding 216 00:14:21,810 --> 00:14:25,380 box and the center coordinates of the bounding box. 217 00:14:25,380 --> 00:14:30,270 So these are the four parameters that are required, uh, that we should pass in. 218 00:14:30,270 --> 00:14:35,010 The input of the Deepsort object tracking algorithm is the center coordinates of the bounding boxes, 219 00:14:35,010 --> 00:14:38,970 and the width and the height, width and the height of the bounding boxes. 220 00:14:38,970 --> 00:14:45,300 So from the Yolo v9 object detection algorithm, we get the uh top left corner, uh top left corner 221 00:14:45,300 --> 00:14:50,280 bounding box coordinates, and the bottom right corner bounding box coordinates. 222 00:14:50,280 --> 00:14:53,130 So this is what we get from the Yolo v9 output. 223 00:14:53,130 --> 00:14:59,790 So we need to transform that output into the height and width of the bounding boxes and the center coordinates 224 00:14:59,790 --> 00:15:02,010 of the bounding boxes. 225 00:15:03,020 --> 00:15:04,490 So first we need to do this. 226 00:15:04,520 --> 00:15:10,130 We need to transform the benign output into this height width and the center coordinates of the bounding 227 00:15:10,130 --> 00:15:10,520 boxes. 228 00:15:11,180 --> 00:15:13,340 So over here I will start writing the code. 229 00:15:22,340 --> 00:15:26,000 So from, um, nine, we get that, uh. 230 00:15:27,130 --> 00:15:31,750 Top left corner bounding box and the bottom right corner bounding box coordinates, which we are here. 231 00:15:31,780 --> 00:15:35,500 Thus, we also get the confidence score and the class names. 232 00:15:39,520 --> 00:15:42,850 So these are the output we get from the YOLO v9. 233 00:15:44,440 --> 00:15:47,470 So here you can just write. 234 00:15:47,470 --> 00:15:50,710 So X1Y1 represents the top left corner. 235 00:15:50,710 --> 00:15:52,840 And X2Y2 represents the bottom right corner. 236 00:15:52,840 --> 00:15:53,980 Bounding box coordinates. 237 00:15:57,690 --> 00:16:00,120 So the output in the form of tensors. 238 00:16:00,120 --> 00:16:02,460 We need to convert the output into integers first. 239 00:16:02,460 --> 00:16:13,680 So I will just write over here int x one, int y one and x two and int y two. 240 00:16:14,430 --> 00:16:17,550 So we need to have transformed the output from tensors into integers. 241 00:16:20,150 --> 00:16:21,950 So now we will find the. 242 00:16:24,000 --> 00:16:25,560 Enter coordinates. 243 00:16:27,530 --> 00:16:29,120 For each of the. 244 00:16:31,280 --> 00:16:33,140 Detected object. 245 00:16:36,840 --> 00:16:42,690 So, as I told you, that in Deepsort algorithm, uh, we need to pass the input in the form of center 246 00:16:42,690 --> 00:16:48,810 coordinates for each of the bounding box of the detected object and the width and height of the bounding 247 00:16:48,810 --> 00:16:49,530 boxes. 248 00:16:50,450 --> 00:16:50,690 Okay. 249 00:16:50,720 --> 00:16:53,450 So we will find the center coordinates first. 250 00:16:53,960 --> 00:16:57,380 And then we'll find the width and the height of the bounding boxes as well. 251 00:16:58,760 --> 00:17:00,620 So I will just ride over here. 252 00:17:17,760 --> 00:17:19,500 So here I will just find the width. 253 00:17:22,940 --> 00:17:24,140 Eye of the. 254 00:17:25,860 --> 00:17:26,220 Only. 255 00:17:38,840 --> 00:17:40,700 So to find the weight. 256 00:17:40,820 --> 00:17:45,350 So you can say that, uh, basically we here we have x1. 257 00:17:45,980 --> 00:17:46,340 Okay. 258 00:17:46,340 --> 00:17:48,980 So if I just write over here. 259 00:17:50,580 --> 00:17:50,760 And. 260 00:17:51,810 --> 00:17:53,070 And X1. 261 00:17:54,220 --> 00:17:56,110 And here we have. 262 00:17:57,470 --> 00:17:58,700 Uh x2. 263 00:17:59,880 --> 00:18:03,150 So if we subtract x one from x two, we get the width. 264 00:18:03,150 --> 00:18:06,690 And if we subtract y one from y two, we get the height. 265 00:18:10,300 --> 00:18:15,550 So I will just I will just write over here x one minus x four. 266 00:18:15,550 --> 00:18:17,410 So this will give me the weight. 267 00:18:17,410 --> 00:18:21,640 And for the height I will just write over here. 268 00:18:30,680 --> 00:18:32,630 For the Deepsort algorithm. 269 00:18:32,630 --> 00:18:35,450 Accept input in the form of. 270 00:18:36,900 --> 00:18:40,170 Center coordinates and width and height of the bounding box. 271 00:18:42,980 --> 00:18:44,390 We'll be right over here. 272 00:18:44,600 --> 00:18:44,870 We're. 273 00:18:53,770 --> 00:18:54,130 Okay. 274 00:18:56,590 --> 00:18:59,440 So now we need to append all these width. 275 00:18:59,440 --> 00:18:59,950 And how? 276 00:18:59,950 --> 00:19:03,940 Because in a frame there will be multiple objects that will be detected. 277 00:19:03,940 --> 00:19:11,230 And so we have uh center coordinates for their multiple objects and width and height or values for multiple 278 00:19:11,230 --> 00:19:16,180 objects that are detected in our video frame in image or in live webcam feed. 279 00:19:19,450 --> 00:19:22,960 So I will just here append dot append. 280 00:19:25,840 --> 00:19:26,500 Okay. 281 00:19:28,460 --> 00:19:33,200 Then I will also append all the confidence values over here as well. 282 00:19:40,270 --> 00:19:44,140 Plus we have also have these class, uh, IDs as well. 283 00:19:44,140 --> 00:19:47,470 For example, the class ID for the person class is zero. 284 00:19:47,470 --> 00:19:49,870 So but these values in the form of tensors. 285 00:19:49,870 --> 00:19:53,860 So we need to transform those values from tensors into integers as well. 286 00:20:05,970 --> 00:20:09,090 And I will append these values over here. 287 00:20:16,450 --> 00:20:17,650 And, uh. 288 00:20:17,650 --> 00:20:19,300 Okay, so that's done. 289 00:20:20,920 --> 00:20:22,240 Then we have. 290 00:20:24,960 --> 00:20:28,080 So now we need to convert these values into tensors again. 291 00:20:50,300 --> 00:20:58,310 And as I told you, that a deepsort algorithm accept, uh, input in the form of, uh, Deepsort algorithm. 292 00:20:58,310 --> 00:21:03,740 Object tracking algorithm in accepts input in the form of, uh, center coordinates and width and height 293 00:21:03,740 --> 00:21:04,760 of the bounding boxes. 294 00:21:04,760 --> 00:21:07,970 And along with the confidence score and the class names. 295 00:21:10,610 --> 00:21:12,560 Uh, content store and the class IDs. 296 00:21:12,560 --> 00:21:15,530 We don't have the class names yet, so. 297 00:21:18,080 --> 00:21:25,520 So as we have transformed our output in the form of tensors and then have confidence score over here 298 00:21:26,060 --> 00:21:26,960 as well. 299 00:21:30,240 --> 00:21:32,430 So here we have the class IDs. 300 00:21:35,190 --> 00:21:40,290 And is IMS represents the current frame in which we will be doing this. 301 00:21:42,960 --> 00:21:48,060 So if we have deductions like if we have some deductions for. 302 00:21:49,010 --> 00:21:51,410 Then we will be doing the tracking as well. 303 00:21:53,920 --> 00:21:54,880 So. 304 00:21:55,360 --> 00:22:01,780 So now you can see over here, uh, in the Deepsort object tracking algorithm, our input, we have 305 00:22:01,780 --> 00:22:07,000 passed the centre coordinates, width and height of the bounding boxes and the confidence score and 306 00:22:07,000 --> 00:22:08,800 the class IDs and the current frame. 307 00:22:08,890 --> 00:22:12,490 So the what output we get from the Deepsort algorithm. 308 00:22:12,490 --> 00:22:15,400 So here in the Deepsort algorithm. 309 00:22:15,400 --> 00:22:17,080 So if I just write. 310 00:22:18,190 --> 00:22:18,790 What year? 311 00:22:19,690 --> 00:22:21,220 Deepsort algorithm. 312 00:22:23,070 --> 00:22:23,640 Output. 313 00:22:24,240 --> 00:22:25,230 So. 314 00:22:29,010 --> 00:22:30,570 So this is what we get. 315 00:22:30,840 --> 00:22:33,390 If I just again zoom it up for the. 316 00:22:57,730 --> 00:22:59,200 Okay, so here we can. 317 00:23:14,700 --> 00:23:18,180 So in the Deepsort algorithm output, we get a. 318 00:23:19,390 --> 00:23:21,070 Uh, the bounding top, right? 319 00:23:21,100 --> 00:23:28,780 Uh, uh, now, this is a top left corner bounding box coordinate, and, uh, the bottom right, uh, 320 00:23:28,780 --> 00:23:29,800 bounding box coordinate. 321 00:23:29,800 --> 00:23:34,930 So Deepsort algorithm output returns the top left corner bounding box coordinate and the bottom right 322 00:23:34,930 --> 00:23:37,810 corner bounding box coordinates, which you can find over here. 323 00:23:37,810 --> 00:23:40,870 So we get the bounding box coordinates in the Deepsort algorithm. 324 00:23:40,870 --> 00:23:45,310 Output a unique ID for each of the detected objects, plus the Deepsort algorithm. 325 00:23:45,310 --> 00:23:46,780 Also return the class ID. 326 00:23:46,780 --> 00:23:49,030 For example, for the person class, it's zero. 327 00:23:49,030 --> 00:23:53,020 So there are three things that we get from the Deepsort algorithm output. 328 00:23:53,020 --> 00:23:54,940 One is the bounding box coordinates. 329 00:23:54,940 --> 00:23:58,360 The second thing is the unique ID for each of the detected object. 330 00:23:58,360 --> 00:23:59,860 This is what we do in object tracking. 331 00:23:59,860 --> 00:24:04,600 In object tracking, we assign a unique ID to each of the detected object, and we also get the class 332 00:24:04,600 --> 00:24:06,700 ID for each of the class. 333 00:24:06,700 --> 00:24:09,640 For example, for person class, the class ID will be zero. 334 00:24:09,640 --> 00:24:11,800 So in the bounding box coordinate we get the. 335 00:24:13,090 --> 00:24:17,590 Top left corner bounding box coordinates and the bottom right corner bounding box coordinates. 336 00:24:20,420 --> 00:24:21,020 Okay. 337 00:24:21,020 --> 00:24:22,130 So. 338 00:24:23,530 --> 00:24:25,780 First thing will I will write over here. 339 00:24:25,780 --> 00:24:28,030 First we get the bounding box coordinate. 340 00:24:28,030 --> 00:24:33,700 We get the top left corner bounding box coordinates and the bounding bottom right corner bounding box 341 00:24:33,700 --> 00:24:34,330 coordinates. 342 00:24:42,420 --> 00:24:47,460 Unless we get the IDs for unique IDs for each of the detected object. 343 00:24:59,130 --> 00:25:04,200 Plus we also get the object ID, for example, for the person class, the object ID zero, and for the 344 00:25:04,200 --> 00:25:05,820 car class it's 1 or 2. 345 00:25:13,480 --> 00:25:19,510 Okay, so now we need to draw those bounding boxes around each of the detected object. 346 00:25:19,510 --> 00:25:23,050 And we need to add the ID uh for this as well. 347 00:25:23,050 --> 00:25:26,650 So I will be creating a function by the name draw dash boxes. 348 00:25:26,650 --> 00:25:27,430 And. 349 00:25:28,680 --> 00:25:34,590 In the drawer dash boxes, as I will be writing the code so that I can draw bounding boxes around each 350 00:25:34,590 --> 00:25:35,550 of the detected object. 351 00:25:35,550 --> 00:25:42,450 Assign a unique ID for each of the object, and write the class name above each of the detected objects. 352 00:25:42,450 --> 00:25:42,990 So. 353 00:25:44,320 --> 00:25:47,290 So in the input I will pass in the bounding box coordinates. 354 00:25:47,890 --> 00:25:48,310 The. 355 00:25:49,360 --> 00:25:52,030 Then I will be, uh, drawing the trails. 356 00:25:52,030 --> 00:25:55,150 So to see what path the object is following. 357 00:25:55,150 --> 00:25:57,160 So to track the path. 358 00:25:58,980 --> 00:26:02,820 Then I will be passing a unique ID for each detected object in the input. 359 00:26:04,650 --> 00:26:08,940 And, uh, object ID, which is the class ID as well. 360 00:26:08,940 --> 00:26:12,480 So the object is basically the class ID like for the person class is zero. 361 00:26:12,480 --> 00:26:14,550 So these will be all the inputs. 362 00:26:14,550 --> 00:26:15,990 So let's write that door. 363 00:26:15,990 --> 00:26:17,190 Dash box is function. 364 00:26:23,030 --> 00:26:25,700 Okay, so you can just get around this. 365 00:26:30,030 --> 00:26:35,070 So here I will be creating a writing down the code for the function draw dash boxes. 366 00:26:36,340 --> 00:26:38,320 The input will be the frame. 367 00:26:41,900 --> 00:26:43,520 The bounding box coordinates. 368 00:26:44,420 --> 00:26:47,150 Dot rail so that we can track the path. 369 00:26:48,630 --> 00:26:50,130 The identities. 370 00:26:52,700 --> 00:26:56,060 So we can initially set these values to none over here. 371 00:26:57,700 --> 00:27:02,470 Then we have the categories as well, so we can set this as none as well. 372 00:27:03,200 --> 00:27:06,290 So categories are basically the class IDs. 373 00:27:07,780 --> 00:27:09,310 Then we have the offset. 374 00:27:09,310 --> 00:27:11,140 We can set this to zero. 375 00:27:12,760 --> 00:27:14,080 So let's try down the code. 376 00:27:14,080 --> 00:27:15,400 Complete code over here. 377 00:27:18,160 --> 00:27:22,630 So let's find out the height and width of the current frame. 378 00:27:30,350 --> 00:27:31,010 Okay. 379 00:27:31,690 --> 00:27:33,550 Then I will write. 380 00:27:34,780 --> 00:27:35,080 Okay. 381 00:27:37,190 --> 00:27:38,900 For a common box. 382 00:27:41,200 --> 00:27:44,620 So now we will look through each of the bounding box coordinates. 383 00:27:44,740 --> 00:27:45,310 Okay. 384 00:27:51,710 --> 00:27:52,220 Oh no. 385 00:27:52,220 --> 00:27:57,200 I'm looping through each of the bounding box coordinates over here. 386 00:28:09,330 --> 00:28:10,320 So. 387 00:28:11,720 --> 00:28:13,580 Um, in the bounding box coordinates. 388 00:28:13,580 --> 00:28:16,490 Let's add some offset value over here. 389 00:28:37,280 --> 00:28:37,520 And. 390 00:28:42,270 --> 00:28:44,490 So here we have defined the offset value over here. 391 00:28:44,490 --> 00:28:49,620 So so now we are looping through each of uh all the uh bounding boxes. 392 00:28:49,620 --> 00:28:54,420 And here we are just getting the coordinates values for each of the bounding boxes. 393 00:28:54,420 --> 00:28:58,020 So we are looking through each of the bounding box coordinates one by one. 394 00:29:00,530 --> 00:29:03,830 For now, we can find the center point of the bounding box. 395 00:29:03,830 --> 00:29:05,810 So find the center point. 396 00:29:06,170 --> 00:29:06,410 The. 397 00:29:27,060 --> 00:29:30,060 So here we are, just finding the center point of the bounding box. 398 00:29:31,800 --> 00:29:33,480 Then over here. 399 00:29:37,690 --> 00:29:43,090 We are calculating the class IDs for each of the detected object. 400 00:29:57,570 --> 00:30:00,180 Then, uh, using the class IDs. 401 00:30:00,180 --> 00:30:04,950 For example, if the there is a person, then we will have the bounding boxes will have this color. 402 00:30:05,310 --> 00:30:07,500 So we will simply write over here. 403 00:30:11,450 --> 00:30:13,610 And here we will pass the class ID. 404 00:30:18,580 --> 00:30:23,530 Then for each of the detected object, we will have a unique ID for each detected object so that we 405 00:30:23,530 --> 00:30:25,330 can track that object using that ID. 406 00:30:25,330 --> 00:30:30,220 So, uh, in that entities, we have the unique ID for each of the detected object. 407 00:30:35,180 --> 00:30:35,630 Um. 408 00:30:50,650 --> 00:30:55,690 So now we will just, uh, draw a rectangle around each of the detected object. 409 00:30:59,060 --> 00:31:05,990 Or you can simply say we are drawing bounding boxes around each of the detected objects. 410 00:31:09,700 --> 00:31:12,430 And the thickness of the bounding boxes will be two. 411 00:31:21,330 --> 00:31:28,020 So now what we will doing is that we will be using that ID so that if the ID is zero, then the there 412 00:31:28,020 --> 00:31:28,980 will be person class. 413 00:31:28,980 --> 00:31:30,360 If the ID is one, then bicycle. 414 00:31:30,360 --> 00:31:32,940 If the ID two is then it belongs to the car class. 415 00:31:33,300 --> 00:31:33,870 So. 416 00:31:35,150 --> 00:31:38,750 So in the category, we have the ID for each of the class. 417 00:31:39,420 --> 00:31:42,150 And using that ID will find the class name. 418 00:31:47,500 --> 00:31:52,720 So in the label, what we will display is that, uh, first I will display the ID. 419 00:31:58,090 --> 00:32:01,510 Then we have the class name over there as well. 420 00:32:04,710 --> 00:32:11,520 But so do things we have in our, uh, uh, above the bounding boxes where we will display the ID unique 421 00:32:11,520 --> 00:32:13,560 ID, and then we have the class name. 422 00:32:13,560 --> 00:32:15,870 We will not be displaying the confidence score over there. 423 00:32:18,440 --> 00:32:20,840 So now we're finding the text size. 424 00:32:33,460 --> 00:32:36,100 Or we want to team up on that 0.5. 425 00:33:06,360 --> 00:33:09,900 Now we will be creating a rectangle above the bounding boxes as well. 426 00:33:21,510 --> 00:33:25,080 And we want the rectangle to be filled, so we'll just write minus one. 427 00:33:27,070 --> 00:33:31,870 And then we will be adding the text in the rectangle which we have created above the bounding box. 428 00:33:32,320 --> 00:33:40,750 In the text, we will be displaying the label which contains the class name and the uh with uh id unique 429 00:33:40,750 --> 00:33:41,830 ID of that object. 430 00:33:44,600 --> 00:33:47,420 Just setting the location of that label. 431 00:33:52,630 --> 00:33:54,070 I mean, the color will be. 432 00:33:55,030 --> 00:33:57,070 White for the text. 433 00:34:04,300 --> 00:34:06,550 And the thickness of the text will be one. 434 00:34:16,440 --> 00:34:22,170 So now we will be creating a circle at the center of each of the object. 435 00:34:40,050 --> 00:34:42,660 Okay, so now we have run this code. 436 00:34:42,660 --> 00:34:46,140 Plus we also want to draw the trails over here as well. 437 00:34:48,420 --> 00:34:51,450 So we can track the path the object is following. 438 00:34:54,180 --> 00:34:57,480 So I will just write down the code for this over here as well. 439 00:35:18,550 --> 00:35:19,030 Okay. 440 00:35:29,120 --> 00:35:29,300 Right. 441 00:35:36,750 --> 00:35:39,930 If I lean not in data deck. 442 00:35:40,020 --> 00:35:43,710 So data deck is a type of dictionary that we have initialized above. 443 00:35:49,190 --> 00:35:51,560 And it has a maximum length of. 444 00:36:00,270 --> 00:36:01,140 64. 445 00:36:26,650 --> 00:36:27,010 Okay. 446 00:36:27,010 --> 00:36:30,460 So now we'll just write over here if Broad Trail. 447 00:36:30,460 --> 00:36:32,620 So if the user wants to draw the trails. 448 00:36:34,330 --> 00:36:36,250 For I in range. 449 00:37:05,850 --> 00:37:06,840 Is none. 450 00:37:25,000 --> 00:37:25,480 Oh, no. 451 00:37:25,480 --> 00:37:27,580 We'll just get greater thickness over here. 452 00:37:32,420 --> 00:37:36,620 So we need to initialize the numpy package at the start over here as well. 453 00:37:37,100 --> 00:37:41,210 So I will just write import numpy as np. 454 00:37:42,380 --> 00:37:44,120 Okay okay. 455 00:37:44,300 --> 00:37:45,170 If okay. 456 00:37:45,170 --> 00:37:46,850 So we have already imported this. 457 00:37:46,850 --> 00:37:48,770 So no need to export that. 458 00:37:48,770 --> 00:37:49,820 Not imported. 459 00:38:03,100 --> 00:38:03,640 Or. 460 00:38:07,400 --> 00:38:10,910 We need to multiply this by 0.5 over here. 461 00:38:14,750 --> 00:38:16,910 Or we can make it 1.5. 462 00:38:21,750 --> 00:38:23,580 Next we need to draw the trail. 463 00:38:23,580 --> 00:38:28,470 So to draw the trails we will be using CV two dot line so that we can draw trails. 464 00:38:29,960 --> 00:38:31,400 The input will reframe. 465 00:38:34,470 --> 00:38:34,770 Okay. 466 00:38:41,540 --> 00:38:42,230 One. 467 00:38:44,310 --> 00:38:46,650 And here we have data dq ID. 468 00:38:48,010 --> 00:38:48,910 One. 469 00:38:50,700 --> 00:38:54,690 Then we have the color and the thickness as our input as well. 470 00:38:54,690 --> 00:38:55,470 And. 471 00:39:03,890 --> 00:39:06,140 So that's quite fine for now. 472 00:39:07,590 --> 00:39:13,470 I think I have written all this, so let me execute this script file and show you how our output looks 473 00:39:13,470 --> 00:39:14,070 like. 474 00:39:16,620 --> 00:39:21,030 Before I run this, I forgot that I have not downloaded, uh, the model weights. 475 00:39:21,960 --> 00:39:27,240 So I will just go over here and just click over here and the model weights will start downloading. 476 00:39:27,600 --> 00:39:33,390 So if I just go over here let me just close these files. 477 00:39:35,010 --> 00:39:35,130 And. 478 00:39:36,250 --> 00:39:40,300 So I should have over here the model which saved. 479 00:39:41,530 --> 00:39:44,290 So if I just check over here. 480 00:39:44,320 --> 00:39:45,310 Okay. 481 00:39:46,870 --> 00:39:48,370 Uh, so let's wait for this model. 482 00:39:48,370 --> 00:39:49,720 Weights to get downloaded. 483 00:39:51,790 --> 00:39:54,100 So it will take a few more seconds for this model. 484 00:39:54,100 --> 00:39:56,110 Weights to get downloaded. 485 00:39:59,560 --> 00:39:59,800 Okay. 486 00:39:59,800 --> 00:40:01,750 So they are about to complete okay. 487 00:40:01,750 --> 00:40:02,560 So that's done. 488 00:40:05,770 --> 00:40:10,480 Okay, so let me just add over here. 489 00:40:14,700 --> 00:40:15,960 So I will just add there. 490 00:40:15,960 --> 00:40:16,620 Over here. 491 00:40:16,650 --> 00:40:18,570 Now you can see it. 492 00:40:20,290 --> 00:40:23,830 Okay, so let me just run this, uh, now in the PyCharm. 493 00:40:23,830 --> 00:40:26,740 So let me just run the script and show you how our output looks like. 494 00:40:29,960 --> 00:40:36,320 Now to run this up over here, we just need to write this Python detect-dual-tracking.py. 495 00:40:36,320 --> 00:40:39,920 Here we I will just define the weights file which we have downloaded. 496 00:40:39,920 --> 00:40:41,870 And that's my input video. 497 00:40:41,870 --> 00:40:45,920 And I'm just running it on CPU because I don't have GPU in my local computer. 498 00:40:45,920 --> 00:40:52,550 So CPU and dash video image so that uh, it can show me the output image or output in the it will open 499 00:40:52,550 --> 00:40:55,880 a window where I can see the output video over here as well. 500 00:40:55,880 --> 00:41:00,770 And my output will be saved into this, uh, folder as well. 501 00:41:00,770 --> 00:41:02,630 So that's all this upload. 502 00:41:06,380 --> 00:41:06,650 Okay. 503 00:41:06,650 --> 00:41:08,480 No such file order. 504 00:41:08,630 --> 00:41:12,770 Okay, so I think mistake is on my side. 505 00:41:12,920 --> 00:41:13,460 Let me. 506 00:41:13,580 --> 00:41:16,490 So we need to set the inline folders current directory. 507 00:41:16,490 --> 00:41:17,720 So now let's run this up. 508 00:41:21,540 --> 00:41:21,720 A. 509 00:41:22,880 --> 00:41:26,720 So this will take few seconds before it starts executing over here. 510 00:41:33,540 --> 00:41:37,680 So here we have the model summary gradient zero parameters. 511 00:41:37,830 --> 00:41:39,540 The parameters are in millions. 512 00:41:39,960 --> 00:41:40,230 Okay. 513 00:41:40,230 --> 00:41:41,760 So you can see the output. 514 00:41:41,760 --> 00:41:43,230 Currently there are no detections. 515 00:41:43,230 --> 00:41:50,610 But uh let's see as we go ahead in the frames uh we have any detection or not. 516 00:41:50,760 --> 00:41:51,120 Okay. 517 00:41:51,120 --> 00:41:53,160 So now you can see here we have the detections. 518 00:41:53,160 --> 00:41:58,290 And you can see over here a unique ID is being assigned to each of the person. 519 00:41:58,530 --> 00:41:59,970 So here we have the person. 520 00:42:00,860 --> 00:42:04,520 Now you can see over here the 19 is put unique IDs for this person. 521 00:42:04,520 --> 00:42:09,410 Two unique IDs for this person 28 unique IDs for this person which disappears. 522 00:42:09,410 --> 00:42:15,320 Five unique IDs for this person 17 unique IDs for this person, 15 unique IDs for this person. 523 00:42:15,320 --> 00:42:17,090 And here we have detected the handbag. 524 00:42:17,090 --> 00:42:20,120 So the 30 unique ID is for this person. 525 00:42:20,120 --> 00:42:26,840 The 14 unique ID is for this person, 12 unique ID is for this person 11 unique IDs for this person, 526 00:42:26,840 --> 00:42:29,210 nine unique ID is for this person. 527 00:42:30,660 --> 00:42:37,710 Now the eight unique IDs for this person, 16 unique IDs for this person, 21 unique IDs for this person. 528 00:42:37,710 --> 00:42:41,550 So each of the person is being assigned a unique ID over here. 529 00:42:41,580 --> 00:42:42,090 Okay. 530 00:42:42,090 --> 00:42:46,950 So if you want to draw the trails and plus you can see we have detected a handbag and we have assigned 531 00:42:46,980 --> 00:42:47,850 ID 30. 532 00:42:47,850 --> 00:42:49,410 So as well. 533 00:42:49,410 --> 00:42:53,850 So if you want to draw the trails uh you can just write over here. 534 00:42:56,150 --> 00:42:58,040 Dash dash, draw, dash trails. 535 00:42:58,040 --> 00:43:04,070 And if you want to select any specific class, like if you don't want to detect Hand-bags in the video, 536 00:43:04,070 --> 00:43:06,290 you can just write classes zero. 537 00:43:06,290 --> 00:43:08,180 So zero is for the person class. 538 00:43:08,180 --> 00:43:12,650 So if you want to know more about the class so you can see zero is for the person, class one is for 539 00:43:12,650 --> 00:43:14,690 the bicycle, class two is for the car. 540 00:43:14,690 --> 00:43:16,640 Class three is for the motorbike class. 541 00:43:16,640 --> 00:43:18,740 So here we have listed 80 different classes. 542 00:43:18,740 --> 00:43:22,370 So it starts from zero and ends at 79 okay. 543 00:43:23,790 --> 00:43:25,260 I won't hear in the classes. 544 00:43:25,260 --> 00:43:29,100 You will assign the ID for which specific class you want to detect. 545 00:43:29,100 --> 00:43:32,910 So if you want to detect all the classes, you can skip the classes over here. 546 00:43:32,910 --> 00:43:37,920 But if you want to detect the specific class, you can just pass out the ID of that specific class or 547 00:43:37,920 --> 00:43:38,730 multiple classes. 548 00:43:38,730 --> 00:43:43,080 You can also create a list and pass the ID of the multiple classes that you want to detect. 549 00:43:44,570 --> 00:43:47,600 So now it will not be returning Hand-bags or any other things. 550 00:43:47,780 --> 00:43:53,030 It will only be detecting the person class because we have passed the ID classes as zero and we have 551 00:43:53,030 --> 00:43:54,860 draw that passed draw trail. 552 00:43:54,860 --> 00:43:57,920 So it will be drawing the trails over here as well. 553 00:44:05,070 --> 00:44:06,420 So let's see now. 554 00:44:06,420 --> 00:44:07,980 Currently there are no detections. 555 00:44:07,980 --> 00:44:13,680 After a few frames initial frames we start to see detections usually okay. 556 00:44:17,350 --> 00:44:20,230 So now you can see over here we have the detections. 557 00:44:20,230 --> 00:44:21,970 Uh okay. 558 00:44:22,600 --> 00:44:22,930 Okay. 559 00:44:22,930 --> 00:44:26,380 So now you can see that red center center as well. 560 00:44:26,380 --> 00:44:28,300 So these are the trails which will not be drawn. 561 00:44:28,300 --> 00:44:32,710 So now you can see as the persons go ahead the trails increase as well. 562 00:44:32,710 --> 00:44:35,890 So you can see over here it's a bit slow. 563 00:44:35,890 --> 00:44:38,500 But now you can see that trails is also increasing. 564 00:44:38,500 --> 00:44:42,220 So these are we are just tracking the path that the person is following. 565 00:44:42,220 --> 00:44:44,320 You can also do it for other classes like car. 566 00:44:44,320 --> 00:44:49,270 You can track the path the car is uh going on and so on. 567 00:44:51,020 --> 00:44:54,200 So now you can see I am drawing the trails over here as well. 568 00:44:54,200 --> 00:44:57,920 We have assign a unique ID using object tracking algorithm. 569 00:44:57,920 --> 00:45:02,510 We are also tracking the path the person is following over here as well. 570 00:45:02,810 --> 00:45:07,160 Okay, so this is all we have done in this code. 571 00:45:07,160 --> 00:45:08,960 We are using object tracking algorithm. 572 00:45:08,960 --> 00:45:12,950 We are tracking each of the detected object and we are drawing the trails over here. 573 00:45:12,950 --> 00:45:16,280 So using trails we are tracking the path like you can see. 574 00:45:16,280 --> 00:45:19,580 So this trails length will increase as the person goes ahead. 575 00:45:19,580 --> 00:45:21,230 You can see over here. 576 00:45:23,020 --> 00:45:29,260 These are the trails that we have drawn, and these trail length will increase as the person goes ahead 577 00:45:29,260 --> 00:45:30,520 further as well. 578 00:45:30,820 --> 00:45:31,510 Okay. 579 00:45:31,510 --> 00:45:33,640 So now you can see over here. 580 00:45:42,550 --> 00:45:46,090 So I have run this script on the Google Colab free GPU as well. 581 00:45:46,090 --> 00:45:46,780 So. 582 00:45:46,780 --> 00:45:50,560 So that I can use the GPU and show you the how fast it works. 583 00:45:50,560 --> 00:45:54,370 So I have the demo video out over here with me as well. 584 00:45:54,370 --> 00:45:57,040 So let me just play that demo video. 585 00:45:57,040 --> 00:46:01,240 So I have run this script on GPU, so you can see the processing comes very fast. 586 00:46:01,240 --> 00:46:06,310 And we have uh, it offers good, uh, frame rate at one GPU. 587 00:46:06,310 --> 00:46:09,970 So you can see we have a good number of detections as well. 588 00:46:09,970 --> 00:46:12,250 And it's the speed is very good as well. 589 00:46:12,250 --> 00:46:14,260 So let me show you the output video over here. 590 00:46:14,260 --> 00:46:19,270 So now you can see over here we are able to detect the person we are, have assigned a unique ID to 591 00:46:19,270 --> 00:46:20,770 each of the person as well. 592 00:46:20,770 --> 00:46:22,270 Plus I'm not drawing the trails. 593 00:46:22,270 --> 00:46:27,790 You can draw the trails if you want as well so that you can track the path the person is following or 594 00:46:27,790 --> 00:46:28,810 any other object. 595 00:46:28,810 --> 00:46:35,020 So in this tutorial we have learned how we can track, uh, each of the detected object, uh, by assigning 596 00:46:35,020 --> 00:46:37,900 a unique ID to each of the objective, uh, detected object. 597 00:46:37,900 --> 00:46:42,580 So in this tutorial we have integrated yolo v9 with Deepsort object tracking algorithm. 598 00:46:42,580 --> 00:46:44,020 And we have, uh. 599 00:46:45,420 --> 00:46:51,780 Attract and detect each of the object as well, and we have assigned a unique ID to each of the object. 600 00:46:51,780 --> 00:46:56,820 Plus we have also, uh, drawn trails to track the path the object is following. 601 00:46:56,820 --> 00:46:58,650 So that's all from this tutorial. 602 00:46:58,650 --> 00:46:59,940 Thank you for watching.