1 00:00:02,000 --> 00:00:03,000 Hello everyone. 2 00:00:03,000 --> 00:00:10,000 In this video tutorial we will see how we can integrate Deepsort object tracking with YOLO. 3 00:00:10,000 --> 00:00:12,000 So this is the complete end to end tutorial. 4 00:00:12,000 --> 00:00:17,000 I will try to explain the complete code of the prediction.py file. 5 00:00:17,000 --> 00:00:18,000 So first of all. 6 00:00:19,000 --> 00:00:20,000 It can start and open. 7 00:00:20,000 --> 00:00:23,000 PyCharm You can use any other for this project. 8 00:00:23,000 --> 00:00:28,000 You can work on Visual Studio, Spyder or using Anaconda Prompt. 9 00:00:28,000 --> 00:00:30,000 You can also implement this project. 10 00:00:30,000 --> 00:00:32,000 So it all depends on your choice. 11 00:00:32,000 --> 00:00:38,000 So here I'm creating a workspace for this project, like you can see here, repairing workspace. 12 00:00:38,000 --> 00:00:42,000 So as you upgrade workspace prepares, I will go to file. 13 00:00:44,000 --> 00:00:45,000 It can create a new pressure. 14 00:00:47,000 --> 00:00:47,000 Thank. 15 00:00:50,000 --> 00:00:55,000 And here I will choose the file object tracking YOLO V8, which is the name of the folder. 16 00:00:56,000 --> 00:00:59,000 And here I will also choose object tracking using YOLO. 17 00:00:59,000 --> 00:01:00,000 So that's fine. 18 00:01:00,000 --> 00:01:03,000 They can create this window. 19 00:01:05,000 --> 00:01:05,000 Okay. 20 00:01:05,000 --> 00:01:08,000 So we have created a new project now. 21 00:01:09,000 --> 00:01:12,000 Well, next I will go to my GitHub repo. 22 00:01:12,000 --> 00:01:14,000 I will share this link with you as well. 23 00:01:14,000 --> 00:01:20,000 So you just need to as this GitHub repo contains the code for your deep sword object tracking. 24 00:01:20,000 --> 00:01:24,000 So we will be using this code to implement this project. 25 00:01:24,000 --> 00:01:25,000 Object tracking. 26 00:01:25,000 --> 00:01:28,000 I will try to explain you the complete code as well. 27 00:01:28,000 --> 00:01:33,000 So first of all, let's run the code while we are doing predictions for a demo video. 28 00:01:33,000 --> 00:01:36,000 In that time I will explain you the complete code as well. 29 00:01:36,000 --> 00:01:39,000 So first of all, let's see the steps to run the code. 30 00:01:41,000 --> 00:01:43,000 So first we need to clone the repository. 31 00:01:43,000 --> 00:01:46,000 Just click on copy over here and let's clone this GitHub repo. 32 00:01:47,000 --> 00:01:48,000 Well. 33 00:01:49,000 --> 00:01:50,000 Yes, exactly. 34 00:01:52,000 --> 00:01:56,000 We'll just paste this over here and just click on Enter. 35 00:01:56,000 --> 00:01:59,000 So now you can see that that rapport is being blown. 36 00:01:59,000 --> 00:02:01,000 So it might take two more seconds. 37 00:02:01,000 --> 00:02:04,000 So let's wait until the rapport gets blown. 38 00:02:05,000 --> 00:02:08,000 Okay, search 67% and now soon it will be hundred percent as well. 39 00:02:12,000 --> 00:02:13,000 75%. 40 00:02:13,000 --> 00:02:14,000 82%. 41 00:02:14,000 --> 00:02:15,000 87%. 42 00:02:15,000 --> 00:02:16,000 100%. 43 00:02:16,000 --> 00:02:19,000 Okay, so now we have clone the GitHub repo over here. 44 00:02:19,000 --> 00:02:22,000 You can see that YOLO V8 Deepsort object tracking. 45 00:02:24,000 --> 00:02:26,000 Now let's look the next step we need to do. 46 00:02:26,000 --> 00:02:30,000 So in the next step, we need to go to the clone repo folder. 47 00:02:30,000 --> 00:02:32,000 So let me show you what is this? 48 00:02:32,000 --> 00:02:39,000 So currently you can see that this project Object tracking YOLO V8 is set to our current directory. 49 00:02:39,000 --> 00:02:46,000 We need to set the current directory as this clone folder name, which is YOLO v8 Deepsort object tracking 50 00:02:46,000 --> 00:02:48,000 should be the name of our current directory. 51 00:02:48,000 --> 00:02:50,000 We need to set this as current directory over. 52 00:02:50,000 --> 00:02:54,000 So just copy and this and paste what we have. 53 00:02:55,000 --> 00:02:57,000 From copied from the GitHub repo. 54 00:02:57,000 --> 00:03:02,000 So now you can see that this folder is being set as our current directory. 55 00:03:02,000 --> 00:03:02,000 Okay. 56 00:03:03,000 --> 00:03:08,000 So in the next step we need to install all the requirements or the dependencies so that we should not 57 00:03:08,000 --> 00:03:14,000 face any error when we run the prediction that this library is not installed or this library is not 58 00:03:14,000 --> 00:03:14,000 installed. 59 00:03:14,000 --> 00:03:21,000 So by running this cell, by running this one inch, install all the required libraries and dependencies. 60 00:03:21,000 --> 00:03:25,000 So just copy this from here and just paste it over here. 61 00:03:25,000 --> 00:03:31,000 So now all the required libraries or the dependencies will be installed as I have run this port previously 62 00:03:31,000 --> 00:03:31,000 as well. 63 00:03:31,000 --> 00:03:34,000 So you can see that requirements are already satisfied. 64 00:03:34,000 --> 00:03:40,000 But if you are running this port for the first time, then you need to install all the libraries. 65 00:03:40,000 --> 00:03:43,000 So the dependencies are now installed. 66 00:03:43,000 --> 00:03:45,000 So guys, as we are doing detection. 67 00:03:45,000 --> 00:03:50,000 So we need to go to that predictions folder inside this repo which is Ultralytics, YOLO. 68 00:03:50,000 --> 00:03:52,000 We hit the deck. 69 00:03:52,000 --> 00:03:57,000 So this is the detection folder which contains the training step validation script and prediction script. 70 00:03:57,000 --> 00:04:04,000 So as in this tutorial I am doing only a predicting uh, on the pre-trained model, so I will be only 71 00:04:04,000 --> 00:04:06,000 using the prediction script. 72 00:04:06,000 --> 00:04:08,000 I will not be using the training or validation step. 73 00:04:10,000 --> 00:04:14,000 So let's set this folder, detect as a word current directory. 74 00:04:14,000 --> 00:04:20,000 So just copy this from here and control V over here. 75 00:04:20,000 --> 00:04:22,000 So this has been set our current directory. 76 00:04:23,000 --> 00:04:28,000 Now we can run the prediction training or validation script where we will only be focusing on the prediction 77 00:04:28,000 --> 00:04:29,000 script. 78 00:04:29,000 --> 00:04:33,000 Next, we need to download the Deepsort files from the Google Drive. 79 00:04:33,000 --> 00:04:39,000 Or just copy this from here and just paste this over here and click on Enter. 80 00:04:40,000 --> 00:04:44,000 So you just need to go to right click over here and. 81 00:04:45,000 --> 00:04:47,000 Okay, taking time. 82 00:04:52,000 --> 00:04:53,000 We just click on download. 83 00:04:54,000 --> 00:04:57,000 So now it first it will zip and then it will download. 84 00:04:57,000 --> 00:04:58,000 It will not take much time. 85 00:04:58,000 --> 00:05:01,000 It will take around 30 seconds to 40 seconds more. 86 00:05:01,000 --> 00:05:05,000 So let's see what it will not be very time taking process. 87 00:05:05,000 --> 00:05:09,000 This the file is zipping and very soon the file has zipped. 88 00:05:09,000 --> 00:05:10,000 Now the download has started. 89 00:05:11,000 --> 00:05:11,000 Okay. 90 00:05:15,000 --> 00:05:16,000 Song. 91 00:05:18,000 --> 00:05:21,000 I could just go to the showing folder. 92 00:05:21,000 --> 00:05:25,000 Just copy this from here and just. 93 00:05:26,000 --> 00:05:28,000 This is our folder which we have created for this project. 94 00:05:28,000 --> 00:05:29,000 I will paste it over here. 95 00:05:30,000 --> 00:05:33,000 I will unzip this first extract files. 96 00:05:36,000 --> 00:05:36,000 Okay. 97 00:05:36,000 --> 00:05:39,000 So now I will get go inside this folder. 98 00:05:40,000 --> 00:05:42,000 Okay, So just copy this from here. 99 00:05:44,000 --> 00:05:53,000 And just go to your deep sort of tracking ultralytics Yolo v it detect and just paste this folder over 100 00:05:53,000 --> 00:05:54,000 here. 101 00:05:54,000 --> 00:05:59,000 This is the Deepsort folder as we are implementing object tracking using Deepsort. 102 00:05:59,000 --> 00:06:02,000 So we need to have that deepsort files. 103 00:06:03,000 --> 00:06:03,000 On our side. 104 00:06:03,000 --> 00:06:08,000 So we have just downloaded the Deepsort files which are necessary to implement object tracking using 105 00:06:08,000 --> 00:06:09,000 Deepsort. 106 00:06:09,000 --> 00:06:14,000 So if we are using a providing object tracking using SORT algorithm, we will require to have the sort 107 00:06:14,000 --> 00:06:15,000 algorithm files. 108 00:06:15,000 --> 00:06:19,000 So here as we are implementing object tracking using deepsort. 109 00:06:19,000 --> 00:06:24,000 So we need to have that deepsort files over here so you can see that Deepsort PyTorch files are over 110 00:06:24,000 --> 00:06:25,000 here. 111 00:06:25,000 --> 00:06:26,000 Okay. 112 00:06:27,000 --> 00:06:28,000 So. 113 00:06:30,000 --> 00:06:36,000 But I have completed this step which is download the Deepsort files from the Google Drive because I'm 114 00:06:36,000 --> 00:06:38,000 implementing object tracking using deepsort. 115 00:06:38,000 --> 00:06:40,000 So I need to add that deepsort files. 116 00:06:40,000 --> 00:06:45,000 So after downloading the deepsort zip file from the drive, unzip it so I've unzipped it and place into 117 00:06:45,000 --> 00:06:47,000 the YOLO detect folder. 118 00:06:47,000 --> 00:06:49,000 So I have completed this step. 119 00:06:49,000 --> 00:06:54,000 So now we need to implement tracking using on some demo video. 120 00:06:54,000 --> 00:06:56,000 So we need to have the demo video as well. 121 00:06:56,000 --> 00:07:00,000 So let's download the demo video by just popping this command. 122 00:07:00,000 --> 00:07:06,000 So I have placed the demo video on my Google Drive, so by just copying and pasting, this command will 123 00:07:06,000 --> 00:07:10,000 download the demo video from the Google Drive into the detect folder. 124 00:07:12,000 --> 00:07:16,000 Just paste this one over here and click on and drag. 125 00:07:16,000 --> 00:07:22,000 So now the demo video will be downloaded like you can see that it will be appearing here. 126 00:07:23,000 --> 00:07:25,000 Let me refresh it so you can see the test. 127 00:07:25,000 --> 00:07:28,000 Dot MP4 is the name of our demo video. 128 00:07:28,000 --> 00:07:33,000 Now let's run the script for YOLO V8 Object detection plus tracking again. 129 00:07:34,000 --> 00:07:39,000 So just copy this over here so you can see that we are using YOLO V8 model over here. 130 00:07:39,000 --> 00:07:45,000 So we have 5 to 6 different modes or sorry, five different models in YOLO, V8, YOLO V8 and is the 131 00:07:45,000 --> 00:07:47,000 smallest, but it is. 132 00:07:47,000 --> 00:07:49,000 It is fast and very less accurate. 133 00:07:49,000 --> 00:07:54,000 Yolo V8X is the largest, so yolo be eight x is more accurate, but it is very less. 134 00:07:54,000 --> 00:07:58,000 Its speed is very less as compared to the other YOLO models. 135 00:07:58,000 --> 00:08:03,000 So we are using a medium because V8 model which is around the other V8 and or you can use your the V8 136 00:08:03,000 --> 00:08:05,000 MPV as well. 137 00:08:05,000 --> 00:08:07,000 Just copy this from here. 138 00:08:08,000 --> 00:08:09,000 And. 139 00:08:10,000 --> 00:08:12,000 Is the script over here and click on Enter. 140 00:08:12,000 --> 00:08:16,000 So here we have passed the name of our demo video file. 141 00:08:16,000 --> 00:08:20,000 Here we have the model name and show is equal to true means. 142 00:08:20,000 --> 00:08:23,000 We want to check the output in the runtime. 143 00:08:23,000 --> 00:08:25,000 Like if show is not is equal to true. 144 00:08:25,000 --> 00:08:31,000 We don't write this, then we are the demo after processing the complete video or after running detection 145 00:08:31,000 --> 00:08:37,000 or tracking on the complete video, it will be saved into my required folder, but I want to see the 146 00:08:37,000 --> 00:08:41,000 output at the runtime so it and show is equal to true. 147 00:08:41,000 --> 00:08:47,000 So this script is now starting to execute, so it might take some time. 148 00:08:47,000 --> 00:08:51,000 So in the meanwhile I will try to explain with the concrete support of prediction. 149 00:08:51,000 --> 00:08:54,000 So now you can see that the script has started to run. 150 00:08:54,000 --> 00:08:59,000 So like you can see that we have a bounding box around the object. 151 00:08:59,000 --> 00:09:06,000 We have a unique ID with each of the object, like you can see two car, five truck. 152 00:09:06,000 --> 00:09:08,000 So this is detecting it wrongly. 153 00:09:08,000 --> 00:09:13,000 But if you use a larger model of your V8, so it will give more accurate results. 154 00:09:13,000 --> 00:09:17,000 So plus you can will be seeing the trails as well. 155 00:09:17,000 --> 00:09:20,000 You can see these trails like you can see this brown line over here. 156 00:09:20,000 --> 00:09:23,000 So you can see that trails and unique ID as well. 157 00:09:23,000 --> 00:09:29,000 So we have implemented object tracking, but let's see the code and see how does this all process is 158 00:09:29,000 --> 00:09:30,000 done. 159 00:09:30,000 --> 00:09:35,000 So let's see the code so you can see that we have the unique ID and the tracking trails with each of 160 00:09:35,000 --> 00:09:37,000 the detected object as well. 161 00:09:37,000 --> 00:09:39,000 So let me show you the import of the code. 162 00:09:39,000 --> 00:09:43,000 In the meanwhile, this script will be run completely. 163 00:09:43,000 --> 00:09:45,000 Then we can look the output as well. 164 00:09:45,000 --> 00:09:45,000 So. 165 00:09:48,000 --> 00:09:51,000 Guys in front of you, you can see the Predict Pi script. 166 00:09:51,000 --> 00:09:55,000 We have executed our demo video on this predict dot pi script. 167 00:09:55,000 --> 00:09:59,000 So let me explain you this whole code in a best way I can. 168 00:09:59,000 --> 00:10:02,000 So first of all, we have imported the Hydra library. 169 00:10:03,000 --> 00:10:05,000 Uh, in the next step, we have got an import torch. 170 00:10:05,000 --> 00:10:09,000 So as it is built using pi dot. 171 00:10:09,000 --> 00:10:17,000 So to perform object detection tracking, using YOLO, we need to import the module so you to use the 172 00:10:17,000 --> 00:10:18,000 PI torch library. 173 00:10:18,000 --> 00:10:20,000 We do import torch. 174 00:10:20,000 --> 00:10:28,000 So to check the version of the PI torch library, we can do print torch dot dash dash version so we 175 00:10:28,000 --> 00:10:31,000 can check the version of the torch library. 176 00:10:31,000 --> 00:10:33,000 So to. 177 00:10:34,000 --> 00:10:38,000 If you face an error in import torch like this module is not available. 178 00:10:38,000 --> 00:10:44,000 So you just need to go to the Chrome store and browser which you have just right pi torch over here 179 00:10:45,000 --> 00:10:48,000 and just go to the pi.site, click on Get Started. 180 00:10:49,000 --> 00:10:53,000 So if this module is not available, you can just go over here pi Torch build. 181 00:10:53,000 --> 00:10:58,000 So you will select the stable, build your operating system which you have, you can select, plus you 182 00:10:58,000 --> 00:11:04,000 can select PIP a language python, and if you have a GPU, you can select you deliver point six 11.7. 183 00:11:04,000 --> 00:11:08,000 But as I don't have a GPU or CPU so I will select on CPU. 184 00:11:08,000 --> 00:11:10,000 So you just need to copy this from here. 185 00:11:11,000 --> 00:11:16,000 Just click on copy and you can just paste this command in your terminal and click on Enter. 186 00:11:16,000 --> 00:11:22,000 So this will install the PI torch library or and when you do import torch, you will not again face 187 00:11:22,000 --> 00:11:23,000 this error. 188 00:11:25,000 --> 00:11:35,000 Next we have import torch dot backends, dot cuda and then so basically torch dot backends control the 189 00:11:35,000 --> 00:11:38,000 behavior of various backends that PyTorch support. 190 00:11:38,000 --> 00:11:44,000 So these backends include torch dot backends, dot dot dot backends, dot u n. 191 00:11:44,000 --> 00:11:45,000 So there are multiple backends. 192 00:11:45,000 --> 00:11:48,000 So bus just have mentioned two over here. 193 00:11:48,000 --> 00:11:51,000 Next we are importing from numpy import random. 194 00:11:51,000 --> 00:11:57,000 So numpy basically random is a module present in the numpy library. 195 00:11:57,000 --> 00:12:02,000 So this module contains the functions which are used for generating random numbers. 196 00:12:02,000 --> 00:12:09,000 So we are using this module to basically randomly generate color palettes for our different detected 197 00:12:09,000 --> 00:12:09,000 objects. 198 00:12:09,000 --> 00:12:12,000 We are not assigning a color palette to each of the objects. 199 00:12:12,000 --> 00:12:15,000 We have assigned a specific color palette to some specific object. 200 00:12:15,000 --> 00:12:22,000 But as we are running this script, our Pre-trained model is trained on coco dataset, so it contains 201 00:12:22,000 --> 00:12:28,000 around 80 classes, so it will randomly select a color palette for each of the object and sign it. 202 00:12:28,000 --> 00:12:32,000 So to randomly select a color better, we have imported the random library. 203 00:12:33,000 --> 00:12:37,000 So here we have importing some required libraries. 204 00:12:37,000 --> 00:12:38,000 Which are these? 205 00:12:38,000 --> 00:12:44,000 Next, as we are implementing object detection and tracking, we need OpenCV Python library, which 206 00:12:44,000 --> 00:12:45,000 is import cv2. 207 00:12:47,000 --> 00:12:50,000 Here we are importing the deepsort PyTorch libraries required. 208 00:12:51,000 --> 00:12:52,000 So. 209 00:12:53,000 --> 00:12:54,000 So these libraries can be. 210 00:12:55,000 --> 00:13:01,000 For example, if you order Deepsort PyTorch folder and you did go to utils and inside the. 211 00:13:02,000 --> 00:13:04,000 parser.py file. 212 00:13:04,000 --> 00:13:12,000 You will find the configuration files and if you go to deepsort or this folder, you will find the Deepsort 213 00:13:12,000 --> 00:13:14,000 folder over here which is deepsort. 214 00:13:14,000 --> 00:13:20,000 So we are importing all these required libraries of the deepsort into our main script file. 215 00:13:20,000 --> 00:13:22,000 predict.py file. 216 00:13:22,000 --> 00:13:27,000 Next we are initializing or we are importing from collections. 217 00:13:27,000 --> 00:13:32,000 So tkeo is a double ended queuing python is implementing using the module collection. 218 00:13:32,000 --> 00:13:33,000 So why? 219 00:13:33,000 --> 00:13:36,000 So you can have a question that why we need the queue. 220 00:13:36,000 --> 00:13:37,000 Why we can't use list. 221 00:13:37,000 --> 00:13:44,000 So for our list in the cases we are need a quicker abandoned operations from the both end of the container. 222 00:13:44,000 --> 00:13:50,000 So here we are using when basically we add an entities into our list. 223 00:13:50,000 --> 00:13:57,000 So when the object disappears from the frame, we need to remove the object ID from that list. 224 00:13:57,000 --> 00:13:59,000 So and and a new object appears in a frame. 225 00:13:59,000 --> 00:14:06,000 We need to add that object ID into the list so for using the queue we can have a quick add and pop operation. 226 00:14:06,000 --> 00:14:11,000 Pop is basically a removal operation, so when we assign a detect an object, we assign it a unique 227 00:14:11,000 --> 00:14:16,000 ID When the object disappears from the frame, we need to remove the ID, and when the new object appears 228 00:14:16,000 --> 00:14:23,000 in the frame, we need to get the we need to assign that ID so to assigning the ID or removing the ID 229 00:14:23,000 --> 00:14:24,000 when the object disappears. 230 00:14:24,000 --> 00:14:28,000 So this is a quick and pop operation which can be done using the queue. 231 00:14:29,000 --> 00:14:33,000 So next, if we want to convert our list into an array, we use a numpy library. 232 00:14:35,000 --> 00:14:40,000 So now we are defining the color palette because we will have a wide variety of tracks because they 233 00:14:40,000 --> 00:14:41,000 are different objects. 234 00:14:41,000 --> 00:14:47,000 So we have establish a color palette over here, so the color of the bounding boxes and the tracking 235 00:14:47,000 --> 00:14:49,000 trails will be the same for each of the objects. 236 00:14:49,000 --> 00:14:55,000 So like you can say if we have cars, so the cars bounding boxes and so on, the cars bounding boxes, 237 00:14:55,000 --> 00:14:59,000 color and train tracking previous color will be the same. 238 00:14:59,000 --> 00:15:04,000 So let me show you this by an image so that I can better explain this point. 239 00:15:07,000 --> 00:15:07,000 What? 240 00:15:07,000 --> 00:15:12,000 I've just taken this image from the output video which is displaying over here. 241 00:15:12,000 --> 00:15:20,000 So if you can see over here, let me pick up and say, no, it's one or something like, no, this is 242 00:15:20,000 --> 00:15:21,000 too much. 243 00:15:22,000 --> 00:15:25,000 That is fine, but there should be some good one. 244 00:15:26,000 --> 00:15:27,000 Okay. 245 00:15:27,000 --> 00:15:29,000 More Saturday night, girl. 246 00:15:32,000 --> 00:15:33,000 Okay, so that's good. 247 00:15:33,000 --> 00:15:36,000 So you can see that on the cars. 248 00:15:36,000 --> 00:15:40,000 You can see over here have a like a light pink color you can say. 249 00:15:40,000 --> 00:15:42,000 And the bounding box color is also pink. 250 00:15:42,000 --> 00:15:48,000 And these trails, like you can see that these trails color is also pink and the bounding box and the 251 00:15:48,000 --> 00:15:54,000 labor label color is also being over here as if you can see that there is a truck. 252 00:15:54,000 --> 00:16:00,000 So you can see that truck bounding box color is brown and this label color is also brown. 253 00:16:00,000 --> 00:16:06,000 So if we have some other object here as well, so its color will also be disciplined and dreams will 254 00:16:06,000 --> 00:16:09,000 be of that same color, which is of the bounding box color. 255 00:16:09,000 --> 00:16:13,000 So for each object like this is that truck and this is a car. 256 00:16:13,000 --> 00:16:18,000 So for each object we have the same bounding box color and the trails color. 257 00:16:18,000 --> 00:16:18,000 Okay. 258 00:16:19,000 --> 00:16:24,000 While if the for example, there is a truck, the truck will have a brown color and its trails will 259 00:16:24,000 --> 00:16:25,000 have the brown color. 260 00:16:25,000 --> 00:16:30,000 If there is a car, the car will have the pink color and its trail will have the also have the pink 261 00:16:30,000 --> 00:16:30,000 color. 262 00:16:30,000 --> 00:16:31,000 Okay. 263 00:16:32,000 --> 00:16:38,000 So here we are defining the color palette because we will have a wide variety of tracks, like we can 264 00:16:38,000 --> 00:16:41,000 have around 80 different objects in a video. 265 00:16:41,000 --> 00:16:43,000 So we want to establish a color pattern. 266 00:16:43,000 --> 00:16:48,000 So the color of the bounding boxes and the tracking frames will be same for each object which we have 267 00:16:48,000 --> 00:16:48,000 seen. 268 00:16:48,000 --> 00:16:54,000 So here we are, initializing a dictionary by the name data, okay? 269 00:16:54,000 --> 00:16:57,000 And here we have just initialized the sword as none. 270 00:16:57,000 --> 00:17:03,000 So here we are, initializing the deepsort and configuration files inside this initialize tracker function. 271 00:17:04,000 --> 00:17:11,000 So inside this track, inside this initialize tracker function, we are initializing the deepsort and 272 00:17:11,000 --> 00:17:13,000 all the required configuration files. 273 00:17:13,000 --> 00:17:18,000 Okay, so here we have a first function define x, y, x, y to x, y w. 274 00:17:18,000 --> 00:17:19,000 So. 275 00:17:20,000 --> 00:17:21,000 So basically. 276 00:17:26,000 --> 00:17:32,000 Because jewelry here gives us the predictions in the form of corner coordinates. 277 00:17:32,000 --> 00:17:38,000 But Deepsort requires the prediction in the form of center coordinates and the width and the height 278 00:17:38,000 --> 00:17:39,000 of the bounding boxes. 279 00:17:39,000 --> 00:17:42,000 Let me explain you this why we need this function. 280 00:17:42,000 --> 00:17:49,000 So, for example, let me just go over here and create a So, for example, this is a bounding box and 281 00:17:49,000 --> 00:17:51,000 here we have an object inside it. 282 00:17:51,000 --> 00:17:53,000 So the object can be anything. 283 00:17:53,000 --> 00:17:54,000 So. 284 00:17:55,000 --> 00:17:58,000 So you can say that this is our object inside the bounding box. 285 00:17:58,000 --> 00:17:59,000 So. 286 00:18:00,000 --> 00:18:03,000 This is our bounding box and this is our object inside the bounding box. 287 00:18:03,000 --> 00:18:03,000 So. 288 00:18:03,000 --> 00:18:07,000 So it will give us prediction that we right over here. 289 00:18:08,000 --> 00:18:10,000 In the form of x1 y1. 290 00:18:11,000 --> 00:18:14,000 Okay, so x1 y1 is basically the. 291 00:18:15,000 --> 00:18:21,000 Top of this is the top corner, 2X1Y1 represent this top, top left corner. 292 00:18:21,000 --> 00:18:22,000 Okay. 293 00:18:22,000 --> 00:18:29,000 So you can in the output we will get this coordinate X1Y1 over here. 294 00:18:29,000 --> 00:18:30,000 What is this? 295 00:18:30,000 --> 00:18:32,000 And this is the bottom right corner. 296 00:18:32,000 --> 00:18:36,000 This value will given by the desert, which is X2Y2. 297 00:18:37,000 --> 00:18:37,000 Okay. 298 00:18:37,000 --> 00:18:45,000 So only these two values will be provided by the model, like the the top left corner and the bottom 299 00:18:45,000 --> 00:18:47,000 right coordinate corner coordinates. 300 00:18:47,000 --> 00:18:53,000 So only these two coordinates values will be provided to us like this. 301 00:18:53,000 --> 00:18:59,000 This coordinate value will be X1Y1 in the form of and this corner value, bottom right corner value 302 00:18:59,000 --> 00:19:01,000 will be in the form of X2Y2. 303 00:19:01,000 --> 00:19:05,000 So these two values will be provided by the YOLO V8 model. 304 00:19:05,000 --> 00:19:11,000 But what Deepsort requires is Deepsort requires the prediction in the form of center coordinates. 305 00:19:11,000 --> 00:19:12,000 Deepsort wants us the. 306 00:19:12,000 --> 00:19:14,000 Let me show you that. 307 00:19:14,000 --> 00:19:21,000 So Deepsort wants us this point these coordinates so X sleeps or wants predictions in the form of x, 308 00:19:21,000 --> 00:19:27,000 y, z mean the center coordinates Deepsort only want the center coordinates and weight. 309 00:19:28,000 --> 00:19:31,000 Weight and height of the bounding box is like, this is the height. 310 00:19:33,000 --> 00:19:36,000 This is the height of the bounding box and the width of the bounding box. 311 00:19:36,000 --> 00:19:37,000 Okay. 312 00:19:37,000 --> 00:19:39,000 So it gives us. 313 00:19:39,000 --> 00:19:42,000 Let me write in a more brighter color. 314 00:19:43,000 --> 00:19:44,000 Okay, so. 315 00:19:45,000 --> 00:19:49,000 Yellowbeard gives us prediction in the form of X1Y1. 316 00:19:49,000 --> 00:19:56,000 This is the top left coordinate over here and the bottom right coordinate so it gives a prediction. 317 00:19:56,000 --> 00:19:59,000 The form of x one, y one and X2Y2. 318 00:19:59,000 --> 00:20:04,000 What Deepsort wants a prediction in the form of center coordinates like x, y, z. 319 00:20:06,000 --> 00:20:12,000 And the width and the height of the bounding boxes like width and height of the bounding boxes. 320 00:20:12,000 --> 00:20:13,000 This is the height. 321 00:20:15,000 --> 00:20:17,000 And this is the width of the bounding boxes. 322 00:20:17,000 --> 00:20:18,000 Okay. 323 00:20:19,000 --> 00:20:21,000 I know I'm writing very poorly. 324 00:20:21,000 --> 00:20:26,000 So in the next step, so after so we have created this function. 325 00:20:26,000 --> 00:20:32,000 This will convert the x1, y1 and x2 y2 coordinates like these to these coordinates. 326 00:20:32,000 --> 00:20:35,000 These coordinates and these coordinates. 327 00:20:35,000 --> 00:20:35,000 So. 328 00:20:37,000 --> 00:20:42,000 So this function will convert these two coordinates into the center, coordinates and width and height 329 00:20:42,000 --> 00:20:44,000 of the bounding boxes. 330 00:20:44,000 --> 00:20:50,000 So as you can see here in the define x one, y, x, y, z, two x, y function, we are converting 331 00:20:50,000 --> 00:20:56,000 the bounding box output received from a V8 to a format that is compatible with deepsort. 332 00:20:56,000 --> 00:21:03,000 Using this function, we convert the x1 x and y coordinates to center coordinates, which is x and y 333 00:21:03,000 --> 00:21:07,000 and return the height and width of the bounding boxes. 334 00:21:07,000 --> 00:21:10,000 So I hope you have realized this importance of this function. 335 00:21:11,000 --> 00:21:15,000 Now we have a function defined x, y, z, 2X2. 336 00:21:15,000 --> 00:21:17,000 And so. 337 00:21:17,000 --> 00:21:21,000 So this function converts the x y coordinates into another format. 338 00:21:21,000 --> 00:21:25,000 In this case, we are only looking for the Doppler width and height. 339 00:21:25,000 --> 00:21:29,000 So this function basically gives us that top left coordinates. 340 00:21:29,000 --> 00:21:30,000 Like let me show you. 341 00:21:32,000 --> 00:21:35,000 So with this second version will only give this coordinate. 342 00:21:36,000 --> 00:21:38,000 The top left coordinate, which is this one. 343 00:21:38,000 --> 00:21:41,000 So only this coordinate will be provided. 344 00:21:41,000 --> 00:21:42,000 Let me show you another color. 345 00:21:42,000 --> 00:21:43,000 So. 346 00:21:44,000 --> 00:21:46,000 Only this coordinate this one. 347 00:21:46,000 --> 00:21:48,000 This point and. 348 00:21:49,000 --> 00:21:51,000 And the width and height of the bounding boxes. 349 00:21:51,000 --> 00:21:55,000 Okay, so this will give us a top left coordinate and width and height of the bounding boxes. 350 00:21:55,000 --> 00:22:00,000 This function will provide but liquid as we are not using this function in this code so you can just 351 00:22:00,000 --> 00:22:01,000 skip this. 352 00:22:01,000 --> 00:22:05,000 So next we are computing colors for the labels. 353 00:22:05,000 --> 00:22:08,000 So here we have defined some colors for the specific labels. 354 00:22:08,000 --> 00:22:12,000 For example, there is a label of person which is zero label. 355 00:22:12,000 --> 00:22:16,000 If we have object that is zero, it will means it represents a class of person. 356 00:22:16,000 --> 00:22:23,000 So if we have a label zero, then we will assign this color palette to the bounding boxes and that fills 357 00:22:23,000 --> 00:22:23,000 or as well. 358 00:22:23,000 --> 00:22:25,000 So these are the trails. 359 00:22:25,000 --> 00:22:28,000 When I'm talking about trails, I'm talking about these lines. 360 00:22:28,000 --> 00:22:29,000 Okay. 361 00:22:29,000 --> 00:22:30,000 These lines. 362 00:22:31,000 --> 00:22:31,000 These lines. 363 00:22:31,000 --> 00:22:32,000 You can see over here. 364 00:22:32,000 --> 00:22:33,000 These lines. 365 00:22:33,000 --> 00:22:34,000 Okay. 366 00:22:34,000 --> 00:22:42,000 So if we have a person object of class, so we'll assign this color to bounding boxes and the trails. 367 00:22:42,000 --> 00:22:44,000 If we have a car, we will assign this color. 368 00:22:44,000 --> 00:22:47,000 So this is light pink color which is being currently being assigned. 369 00:22:47,000 --> 00:22:48,000 This is light in color. 370 00:22:49,000 --> 00:22:53,000 And if the your motorbike is color will decide if you have a bus, this color will be a. 371 00:22:54,000 --> 00:23:01,000 So if the detected object does not belongs to any of these labels or class, then it will randomly select 372 00:23:01,000 --> 00:23:08,000 the color palette and it will show if the detected object is not from any of this class, then it will 373 00:23:08,000 --> 00:23:12,000 randomly select a color and it will assign a in. 374 00:23:12,000 --> 00:23:14,000 The output will get the attribute of that color. 375 00:23:14,000 --> 00:23:15,000 Okay, so. 376 00:23:16,000 --> 00:23:21,000 Next we have our draw our border function in the draw border function. 377 00:23:25,000 --> 00:23:27,000 In draw dash border function. 378 00:23:27,000 --> 00:23:31,000 We have written the script to create a rounded rectangle for the label. 379 00:23:31,000 --> 00:23:32,000 Let me show you. 380 00:23:32,000 --> 00:23:33,000 So. 381 00:23:34,000 --> 00:23:41,000 So you can see over here, this is a rounded rectangle in which contains the 86 cent card. 382 00:23:41,000 --> 00:23:45,000 So to create this rounded rectangle is rounded rectangle only. 383 00:23:46,000 --> 00:23:49,000 I have written the script for the Draw Dash border. 384 00:23:49,000 --> 00:23:55,000 So this draw dash border contains a square with the help of which we can create this rounded rectangle. 385 00:23:55,000 --> 00:23:59,000 Okay, so next we have the dashboard function. 386 00:24:00,000 --> 00:24:04,000 So in the box function we have the cv2 dot rectangle. 387 00:24:04,000 --> 00:24:11,000 So Cv2 dot rectangles create a bounding box like this bounding box around the detected objects like 388 00:24:11,000 --> 00:24:13,000 you can see here as well. 389 00:24:13,000 --> 00:24:16,000 So you can see here as well. 390 00:24:16,000 --> 00:24:16,000 So. 391 00:24:17,000 --> 00:24:19,000 Uh, see, we're using cv2 dot rectangle. 392 00:24:19,000 --> 00:24:23,000 We can create a bounding box around the detected object. 393 00:24:23,000 --> 00:24:23,000 Okay. 394 00:24:24,000 --> 00:24:27,000 And C1 C2 contain the center coordinates. 395 00:24:27,000 --> 00:24:28,000 Okay. 396 00:24:28,000 --> 00:24:33,000 So in the next step, we are calling the draw dash border function over here. 397 00:24:33,000 --> 00:24:36,000 So this is the draw dash border function which I have explained previously. 398 00:24:36,000 --> 00:24:43,000 So now we are calling draw dash border function over here to create a rounded rectangle like this rounded 399 00:24:43,000 --> 00:24:43,000 rectangle. 400 00:24:43,000 --> 00:24:45,000 And this one I've already explained. 401 00:24:45,000 --> 00:24:50,000 And using cv2 dot put text using this Cv2 dot put text. 402 00:24:50,000 --> 00:24:56,000 We are adding the text text in this drawer dash rectangle which is 86 and car. 403 00:24:57,000 --> 00:24:59,000 We'll draw this border. 404 00:24:59,000 --> 00:25:03,000 We create a rounded rectangle like this one. 405 00:25:04,000 --> 00:25:06,000 And using cv2 dot put text. 406 00:25:06,000 --> 00:25:12,000 We add this text like identity ID number or unique ID number of each object and the object name. 407 00:25:13,000 --> 00:25:15,000 Okay, so this is on. 408 00:25:15,000 --> 00:25:19,000 Now we have a draw boxes function in draw box six function. 409 00:25:19,000 --> 00:25:21,000 I am calling the dash box function. 410 00:25:21,000 --> 00:25:22,000 Let me show you its called. 411 00:25:23,000 --> 00:25:24,000 So. 412 00:25:26,000 --> 00:25:28,000 And let me see where it is called. 413 00:25:36,000 --> 00:25:36,000 For him. 414 00:25:36,000 --> 00:25:40,000 The draw dice boxes function we have called the box function. 415 00:25:40,000 --> 00:25:48,000 So the box function will add the rectangle around the y function. 416 00:25:48,000 --> 00:25:53,000 Box function will create this rectangle around the detected object like this rectangle square rectangle 417 00:25:53,000 --> 00:25:59,000 and also add the this rounded rectangle as well and put the text into this as well. 418 00:25:59,000 --> 00:26:02,000 So box function do thing three things. 419 00:26:02,000 --> 00:26:04,000 First, it create a rounded rectangle. 420 00:26:04,000 --> 00:26:08,000 It creates a rectangle around the detected object, which is a square rectangle. 421 00:26:08,000 --> 00:26:14,000 Second, it adds a it creates a rounded rectangle using draw dash border function because we are calling 422 00:26:14,000 --> 00:26:20,000 draw dash border function inside the dash box function plus it assigns a unique ID and the label name. 423 00:26:20,000 --> 00:26:22,000 Okay, so. 424 00:26:24,000 --> 00:26:25,000 Let's go up. 425 00:26:25,000 --> 00:26:31,000 So to draw the bounding boxes around the detected object, assigning unique IDs to each detected object, 426 00:26:31,000 --> 00:26:34,000 assigning labels will be done using the. 427 00:26:35,000 --> 00:26:40,000 The dash box function which we have created over here. 428 00:26:40,000 --> 00:26:46,000 And we are just calling this dash function function in this drawer, dash boxes function and this drawer 429 00:26:46,000 --> 00:26:47,000 dash drawing. 430 00:26:47,000 --> 00:26:50,000 Drawing will also draw the trails for tracking. 431 00:26:50,000 --> 00:26:56,000 So drawing the trails for tracking script will be written in this over here, like this is the trails. 432 00:26:56,000 --> 00:26:57,000 These are the trails. 433 00:26:57,000 --> 00:26:59,000 These are the trails I'm talking about. 434 00:27:00,000 --> 00:27:01,000 Okay. 435 00:27:01,000 --> 00:27:07,000 So as we are doing detection frame by frame, so here I am checking the height and width of the current 436 00:27:07,000 --> 00:27:07,000 frame. 437 00:27:07,000 --> 00:27:08,000 Okay. 438 00:27:08,000 --> 00:27:11,000 So here I'm just checking the height and width of the current frame. 439 00:27:11,000 --> 00:27:12,000 Okay. 440 00:27:12,000 --> 00:27:17,000 So we will store the unique IDs of the detected object until the object is in the frame. 441 00:27:17,000 --> 00:27:19,000 Okay, so let me explain this. 442 00:27:19,000 --> 00:27:25,000 You can see that each object is being assigned a unique ID 87, 53, 67, 68. 443 00:27:25,000 --> 00:27:28,000 So all these IDs, unique IDs. 444 00:27:31,000 --> 00:27:34,000 Basically all these unique IDs will be stored in a list. 445 00:27:34,000 --> 00:27:38,000 Okay, So these will be stored in that list which we have created. 446 00:27:38,000 --> 00:27:43,000 Like we have told you in the start, that using we can add and remove an item from the list. 447 00:27:44,000 --> 00:27:46,000 Okay, so these will be stored in the list. 448 00:27:46,000 --> 00:27:51,000 So until this specific object is in the frame, the ID will be assigned to that object. 449 00:27:51,000 --> 00:27:56,000 When the object is disappears, for example, the yellow card is in the current frame, but after few 450 00:27:56,000 --> 00:28:00,000 frames, the yellow card disappears from the frame. 451 00:28:00,000 --> 00:28:06,000 Okay, so we will remove As the object disappears from the frame, we will remove the unique ID of that 452 00:28:06,000 --> 00:28:09,000 object from the list and will also end. 453 00:28:09,000 --> 00:28:14,000 For example, the yellow card disappears and the black card enters into that frame in our frame. 454 00:28:14,000 --> 00:28:21,000 So we will assign the unique ID to the new object and we will also save the new unique ID of the new 455 00:28:21,000 --> 00:28:22,000 object appearing in the frame. 456 00:28:22,000 --> 00:28:25,000 Okay, so here we have written a script. 457 00:28:25,000 --> 00:28:29,000 So if the object disappears from the frame, we will remove that object. 458 00:28:29,000 --> 00:28:31,000 Unique id. 459 00:28:31,000 --> 00:28:31,000 Okay. 460 00:28:31,000 --> 00:28:38,000 Using dot pop key, we remove the ID of the object from the list if the object is no more in the frame. 461 00:28:39,000 --> 00:28:39,000 Okay. 462 00:28:39,000 --> 00:28:40,000 So next we will. 463 00:28:43,000 --> 00:28:46,000 Well, now we will loop through the bounding boxes one by one. 464 00:28:46,000 --> 00:28:51,000 So here we have the on the paw coordinates for each of the bounding box. 465 00:28:52,000 --> 00:29:00,000 So x one and y one represents the top left corner and X2 and Y2 represent the bottom right corner of 466 00:29:00,000 --> 00:29:01,000 the bounding box. 467 00:29:01,000 --> 00:29:02,000 Let me show you. 468 00:29:03,000 --> 00:29:04,000 I have ordered his plane. 469 00:29:04,000 --> 00:29:06,000 But let me explain this as well. 470 00:29:07,000 --> 00:29:14,000 So if we just clean this up it so that I can in a better position to explain this. 471 00:29:14,000 --> 00:29:15,000 Okay. 472 00:29:15,000 --> 00:29:15,000 So. 473 00:29:15,000 --> 00:29:16,000 Okay. 474 00:29:16,000 --> 00:29:16,000 Okay. 475 00:29:16,000 --> 00:29:18,000 So let me show you. 476 00:29:19,000 --> 00:29:24,000 So we have you can see that multiple bounding boxes over here like one, two, three, four, five, 477 00:29:24,000 --> 00:29:26,000 six, seven, eight, nine. 478 00:29:26,000 --> 00:29:29,000 Okay, so one thing is. 479 00:29:32,000 --> 00:29:33,000 This is basically x one. 480 00:29:33,000 --> 00:29:34,000 Y one. 481 00:29:35,000 --> 00:29:40,000 It is the top left corner and this is the bottom right corner and this represents x2 y2. 482 00:29:40,000 --> 00:29:41,000 Okay. 483 00:29:41,000 --> 00:29:44,000 So for each of the bounding box over here. 484 00:29:45,000 --> 00:29:46,000 Like this one. 485 00:29:46,000 --> 00:29:47,000 This one. 486 00:29:47,000 --> 00:29:47,000 This one. 487 00:29:48,000 --> 00:29:48,000 Okay. 488 00:29:48,000 --> 00:29:53,000 Like for each of the bounding box over here, we have the four coordinates for each of the bounding 489 00:29:53,000 --> 00:29:53,000 box. 490 00:29:55,000 --> 00:29:58,000 Which is x1, y1, x2 and Y2. 491 00:29:58,000 --> 00:30:04,000 Okay, so next we need to find the center of the bottom edge of the bounding box. 492 00:30:05,000 --> 00:30:06,000 So let me show you. 493 00:30:06,000 --> 00:30:10,000 So in the next step, basically, this is the center of the bottom edge of the bounding box. 494 00:30:10,000 --> 00:30:14,000 Like this is the center of the bottom edge of the bounding box. 495 00:30:14,000 --> 00:30:18,000 We need to find the center because we will draw trails from the center. 496 00:30:18,000 --> 00:30:20,000 You can see that these are the trails. 497 00:30:20,000 --> 00:30:21,000 So we need to find the center. 498 00:30:22,000 --> 00:30:26,000 So here we have written the code to find the center of the bottom edge of the bounding box. 499 00:30:28,000 --> 00:30:29,000 Next. 500 00:30:29,000 --> 00:30:29,000 We will. 501 00:30:30,000 --> 00:30:33,000 We need to have a unique ID for each of the project. 502 00:30:33,000 --> 00:30:39,000 So using this script, we'll get a unique ID for each of the detected object. 503 00:30:41,000 --> 00:30:45,000 Okay, so next, we can see if the new object appears in the frame. 504 00:30:45,000 --> 00:30:51,000 So here we have assigned a new ID if the a new objects appear in the frame. 505 00:30:51,000 --> 00:30:57,000 Okay, so now we will create a new buffer or a new list which has the maximum length 64. 506 00:30:57,000 --> 00:30:59,000 So if so, let me show you. 507 00:30:59,000 --> 00:31:03,000 So we have we have created a list which maximum length is 64. 508 00:31:03,000 --> 00:31:08,000 So what we will do here is basically you can see this. 509 00:31:08,000 --> 00:31:10,000 We have drawn the trails over here. 510 00:31:10,000 --> 00:31:18,000 So we have saved this center points basically after getting the center point of this, like after calculating 511 00:31:18,000 --> 00:31:20,000 the bottom center. 512 00:31:20,000 --> 00:31:22,000 Let me try to explain this better way. 513 00:31:24,000 --> 00:31:28,000 So after finding the center of the bottom edge of the bounding box. 514 00:31:28,000 --> 00:31:35,000 So after I find the center of the bottom edge of the bounding box, I will save all the center points 515 00:31:35,000 --> 00:31:36,000 of each of the frame. 516 00:31:36,000 --> 00:31:38,000 Like there are 100 frames in a video. 517 00:31:38,000 --> 00:31:45,000 I will save this center of the bottom edge of the bounding box of each of the each of sorry, of the 518 00:31:45,000 --> 00:31:49,000 bottom edge of the bounding box of each of the frame in a list. 519 00:31:49,000 --> 00:31:57,000 So basically the center of the bottom edge of the bounding box of all the hundred frames in which the 520 00:31:57,000 --> 00:32:00,000 object is appearing is saved in a list. 521 00:32:00,000 --> 00:32:02,000 Okay, which maximum length is 64. 522 00:32:02,000 --> 00:32:10,000 So if a 65 fifth of the object appears, element appears it will remove the first and it will add the 523 00:32:10,000 --> 00:32:11,000 65 into it. 524 00:32:11,000 --> 00:32:12,000 So. 525 00:32:13,000 --> 00:32:18,000 As the new elements appear that all the previous elements will be removed. 526 00:32:18,000 --> 00:32:23,000 Okay, so if we set this line to 128 so one thing will happen. 527 00:32:23,000 --> 00:32:32,000 This trains will go further long because after we save the center coordinates like the bottom edge, 528 00:32:32,000 --> 00:32:40,000 bottom of the the bounding box coordinate and we draw the trails using these coordinates, which we 529 00:32:40,000 --> 00:32:40,000 have saved. 530 00:32:40,000 --> 00:32:42,000 Okay, so. 531 00:32:44,000 --> 00:32:49,000 After saving the center of the of the module, the bounding box coordinates, we draw trees, which 532 00:32:49,000 --> 00:32:50,000 maximum length is 64. 533 00:32:50,000 --> 00:32:55,000 So if we change the maximum length to 128, this change length will go further wrong. 534 00:32:55,000 --> 00:33:00,000 Okay, so now we are setting a unique color for each bounding box and rounding rectangles which could 535 00:33:00,000 --> 00:33:01,000 be in the labor. 536 00:33:02,000 --> 00:33:08,000 Okay, so next we have the object id object id contains the ID for the each of the object, like for 537 00:33:08,000 --> 00:33:10,000 the person, the object ID will be zero. 538 00:33:10,000 --> 00:33:12,000 So when we write names, zero. 539 00:33:12,000 --> 00:33:15,000 So in the names we have the names for on the object. 540 00:33:15,000 --> 00:33:18,000 So if we object, ID is zero over here. 541 00:33:18,000 --> 00:33:20,000 So if there is a zero over here. 542 00:33:20,000 --> 00:33:22,000 So it means that the object name is person. 543 00:33:22,000 --> 00:33:25,000 If there is a one, then it means there is a car or a bus. 544 00:33:25,000 --> 00:33:26,000 Okay. 545 00:33:26,000 --> 00:33:29,000 So now here we are just putting the labels in a required format. 546 00:33:31,000 --> 00:33:38,000 So you can see that we are appending the center points of the bottom edge of the bounding box like we 547 00:33:38,000 --> 00:33:45,000 are appending these center coordinates into our data dequeue list so that a data list is maximum. 548 00:33:45,000 --> 00:33:46,000 That is 64. 549 00:33:46,000 --> 00:33:46,000 So. 550 00:33:47,000 --> 00:33:51,000 Using this center coordinates would be saved in the data list. 551 00:33:51,000 --> 00:33:54,000 And using these coordinates, we will draw the trails. 552 00:33:54,000 --> 00:33:59,000 Okay, so now we are just using the foot box function to create a rounded rectangle. 553 00:34:00,000 --> 00:34:06,000 To create this rounded rectangle and to add the text and to draw this rounded rectangle as well. 554 00:34:06,000 --> 00:34:07,000 To create this rectangle. 555 00:34:07,000 --> 00:34:08,000 Sorry, this is not the rounded. 556 00:34:08,000 --> 00:34:15,000 We are using this box function to create this rectangle and this rounded rectangle and to put that text 557 00:34:15,000 --> 00:34:17,000 over here as well, which is ID and label. 558 00:34:19,000 --> 00:34:23,000 Okay, so now here we are, just drawing the trails. 559 00:34:23,000 --> 00:34:27,000 You can you can see that we are drawing the trails using Cv2 dot nine. 560 00:34:27,000 --> 00:34:33,000 So it basically loops through the data data center points and draw a line between the current and last 561 00:34:33,000 --> 00:34:33,000 point. 562 00:34:33,000 --> 00:34:37,000 So if, if the lattice maximum length is 64. 563 00:34:38,000 --> 00:34:38,000 Okay. 564 00:34:38,000 --> 00:34:46,000 So we have saved a 64 and the center points in the list, if the net is 64, we saved us 64 center points 565 00:34:46,000 --> 00:34:54,000 in the list is 64 and we have the length of 64 trails of 64 center point if we change the maximum length 566 00:34:54,000 --> 00:34:55,000 to 128. 567 00:34:55,000 --> 00:34:56,000 Okay. 568 00:34:56,000 --> 00:34:59,000 So this trails that will go even further. 569 00:34:59,000 --> 00:35:00,000 Okay. 570 00:35:00,000 --> 00:35:07,000 So in this way, all this process is implemented here, we will draw the line of the trails in next. 571 00:35:07,000 --> 00:35:08,000 This is all the code for the. 572 00:35:10,000 --> 00:35:12,000 If we go further. 573 00:35:13,000 --> 00:35:20,000 So what we are doing here, we are just looping over all the detections because it gives us the corner 574 00:35:20,000 --> 00:35:26,000 coordinates like x1, Y1, x2 and Y2, but Deepsort requires the center coordinates and the height and 575 00:35:26,000 --> 00:35:27,000 the width. 576 00:35:27,000 --> 00:35:32,000 So here we are, just converting the corner coordinates like this. 577 00:35:32,000 --> 00:35:33,000 Let me show you. 578 00:35:33,000 --> 00:35:38,000 Here we are converting the coordinate coordinates X1Y1X2Y2. 579 00:35:39,000 --> 00:35:40,000 Those center coordinates. 580 00:35:40,000 --> 00:35:46,000 So we are just want to get the center coordinates and the width and the height of the bounding boxes. 581 00:35:47,000 --> 00:35:49,000 Okay, so. 582 00:35:50,000 --> 00:35:55,000 Using this function, we just pass the coordinate coordinates and get the center coordinates, X, y, 583 00:35:55,000 --> 00:35:58,000 z and width and height of the bounding box. 584 00:35:59,000 --> 00:36:05,000 So now we are using a deepsort update function that is, and this function is called the update function. 585 00:36:05,000 --> 00:36:10,000 The bot takes all the texture and just add the identities because he like you wallaby it will do the 586 00:36:10,000 --> 00:36:11,000 detections. 587 00:36:12,000 --> 00:36:17,000 Deepsort will only assign a unique ID to each of the object and draw the trails. 588 00:36:18,000 --> 00:36:22,000 So dinshaw short, just assign unique ID so these IDs are not incremental. 589 00:36:22,000 --> 00:36:24,000 So what does it means that. 590 00:36:25,000 --> 00:36:31,000 Like you can see that 87, 68, 53, 67 like it is not 87, 88, 89, 90. 591 00:36:31,000 --> 00:36:38,000 It assigns a ID randomly so it can assign a one ID to this object to 1011. 592 00:36:38,000 --> 00:36:40,000 So it is not incremental form. 593 00:36:41,000 --> 00:36:41,000 Okay. 594 00:36:42,000 --> 00:36:44,000 Well, I think this is all from the code. 595 00:36:44,000 --> 00:36:46,000 I hope you have understood this. 596 00:36:46,000 --> 00:36:51,000 In the meanwhile, our script has been dropped, so let me show the output demo video which will be 597 00:36:51,000 --> 00:36:51,000 saved. 598 00:36:52,000 --> 00:36:54,000 Um, in this folder. 599 00:36:54,000 --> 00:36:56,000 Let me go to this folder. 600 00:37:00,000 --> 00:37:02,000 But what was the name of the project? 601 00:37:02,000 --> 00:37:02,000 I think. 602 00:37:04,000 --> 00:37:05,000 Rectangularly eared. 603 00:37:05,000 --> 00:37:09,000 And then we will go to your deepsort alternatives. 604 00:37:09,000 --> 00:37:10,000 Yolo. 605 00:37:11,000 --> 00:37:12,000 We ate. 606 00:37:12,000 --> 00:37:16,000 It runs with the tree. 607 00:37:16,000 --> 00:37:18,000 And here is our output video. 608 00:37:18,000 --> 00:37:24,000 Let's see the results so you can see that we have a unique ID with each of the object. 609 00:37:24,000 --> 00:37:25,000 Thus we have a tracking trace. 610 00:37:26,000 --> 00:37:28,000 We have detecting the object. 611 00:37:28,000 --> 00:37:29,000 We are able to track the object. 612 00:37:30,000 --> 00:37:32,000 Well, the model is working quite fine. 613 00:37:32,000 --> 00:37:36,000 We are able to detect, track and assign a unique ID to each of the object. 614 00:37:37,000 --> 00:37:39,000 So this is all from this video tutorial. 615 00:37:39,000 --> 00:37:41,000 I hope you have learned a lot from it. 616 00:37:41,000 --> 00:37:43,000 See you all in the next video tutorial. 617 00:37:43,000 --> 00:37:44,000 Till then, bye bye.