1 00:00:02,330 --> 00:00:08,750 This video tutorial we will learn how we can do automatic number plate recognition using YOLO v ten 2 00:00:08,750 --> 00:00:10,580 and paddle OCR. 3 00:00:10,610 --> 00:00:17,420 We will also explore how we can save the license plate detection data to a SQL database. 4 00:00:17,420 --> 00:00:19,340 So let's get started with it. 5 00:00:19,340 --> 00:00:24,110 So I've divided this complete tutorial into these five steps which I have mentioned over here. 6 00:00:24,110 --> 00:00:31,380 So first of all we will see how we can detect license plates using YOLO YOLO is basically the latest 7 00:00:31,380 --> 00:00:36,900 state of the art object detection model, and we will be using this YOLO model in this project. 8 00:00:36,900 --> 00:00:41,040 So in the step number one we will detect license plates using yolo v ten. 9 00:00:41,040 --> 00:00:46,020 And in the step number two we will recognize text on the license plate. 10 00:00:46,020 --> 00:00:50,880 Like we will read the text on the license plate like one, two, three or whatever it is. 11 00:00:50,880 --> 00:00:55,060 So we will read the text on the license plate using paddle OCR. 12 00:00:55,840 --> 00:01:00,730 Then in the step number three we will accumulate the unique license plates. 13 00:01:00,730 --> 00:01:06,910 Like we will accumulate the unique license plate means like we will only we will create a list. 14 00:01:06,910 --> 00:01:13,210 We will, uh, add only the unique license plates like there would be no duplications over there. 15 00:01:13,420 --> 00:01:19,970 So then after accumulating the unique license plate, then we'll save the accumulated unique license 16 00:01:19,970 --> 00:01:24,290 plate to a JSON file after every 20s. 17 00:01:25,610 --> 00:01:29,150 So first of all, we will accumulate the unique license plate. 18 00:01:29,150 --> 00:01:35,210 Then we'll save the accumulated unique license plate to a JSON file after every 20s. 19 00:01:35,210 --> 00:01:38,990 So before uh, in the step number two, we will do two things. 20 00:01:38,990 --> 00:01:47,310 So we will after every 20s interval, we will generate a JSON file, uh, in our JSON directory, which 21 00:01:47,310 --> 00:01:48,960 will contain the start time. 22 00:01:48,960 --> 00:01:53,610 The end time, like the 20 interval start time and 22nd interval end time. 23 00:01:53,610 --> 00:01:59,970 And we will have the the directory license plates data which the license plates which are detected during 24 00:01:59,970 --> 00:02:01,320 that 20s interval. 25 00:02:01,320 --> 00:02:08,440 So after every 22nd inverter well we will generate a JSON file which will have the three things. 26 00:02:08,440 --> 00:02:15,430 One is the start time and one is the end time, and the next will be the license plates which will are 27 00:02:15,430 --> 00:02:17,650 detected during this 20s interval. 28 00:02:17,650 --> 00:02:19,510 So there can be five lessons played. 29 00:02:19,510 --> 00:02:24,730 There can be ten lessons played, or there can be only one license plate that is detected during the 30 00:02:24,730 --> 00:02:29,540 22nd interval, or might be the license plate is directly during that 20s interval. 31 00:02:29,540 --> 00:02:31,490 So this is the one thing which we will do. 32 00:02:31,490 --> 00:02:37,550 And the second thing will be that we will create a cumulative JSON file like, um, one thing I told 33 00:02:37,550 --> 00:02:42,170 you in the first step, like, uh, previously like we will be generating a new JSON file after every 34 00:02:42,170 --> 00:02:43,340 20s interval. 35 00:02:43,520 --> 00:02:51,050 Plus we will also generate a one single cumulative JSON file which will contain all the detected license 36 00:02:51,050 --> 00:02:53,220 plate data for that the 20s. 37 00:02:53,520 --> 00:02:54,390 Like this is the. 38 00:02:54,600 --> 00:02:57,930 These are the license plate detected in the first 22nd interval. 39 00:02:57,930 --> 00:03:00,930 These are the license plate detected during the next 22nd interval. 40 00:03:00,960 --> 00:03:04,050 These are the license plate detected during the third 22nd interval. 41 00:03:04,260 --> 00:03:10,740 So in the cumulative JSON file one JSON file will have all the license plate detected uh, during all 42 00:03:10,740 --> 00:03:12,210 the 20s interval. 43 00:03:12,240 --> 00:03:12,840 Okay. 44 00:03:14,640 --> 00:03:19,420 Then in the step number five, what we'll do, we will save the license plate detection data to a SQL 45 00:03:19,420 --> 00:03:20,200 database. 46 00:03:20,200 --> 00:03:26,290 So we will be using SQLite which is a lightweight disk based database and which is easy to set up and 47 00:03:26,290 --> 00:03:26,860 use. 48 00:03:26,860 --> 00:03:33,040 So uh, before we go ahead, let me show you how our simulated JSON file will look like. 49 00:03:33,040 --> 00:03:34,750 So you have no doubts about it. 50 00:03:38,380 --> 00:03:46,130 So now over here you can see that these are all the JSON files that are generated after every 22nd interval. 51 00:03:46,130 --> 00:03:47,480 Like you can see over here. 52 00:03:47,480 --> 00:03:52,100 All these are the JSON files that are generated after every 22nd interval. 53 00:03:52,100 --> 00:03:58,370 And here you can see at the top this is our cumulative JSON file which is given over here by the name 54 00:03:58,370 --> 00:03:59,270 license plates. 55 00:03:59,270 --> 00:04:02,690 So let me open it over here and let me navigate. 56 00:04:03,920 --> 00:04:07,590 So now you can see over here this is the cumulative JSON file. 57 00:04:07,590 --> 00:04:11,070 So now you can see that 2413 2433. 58 00:04:11,070 --> 00:04:12,720 So these are the 22nd table. 59 00:04:12,720 --> 00:04:18,270 And these is the list of the license plate that are detected during that 22nd interval. 60 00:04:18,390 --> 00:04:21,630 So now as the first 22nd interval and over here. 61 00:04:21,630 --> 00:04:24,900 So the next 22nd interval will be starting from here as well. 62 00:04:24,900 --> 00:04:26,010 Like you can see. 63 00:04:26,010 --> 00:04:28,290 And it will end at 2454. 64 00:04:28,290 --> 00:04:33,280 And these are the lists of the license plates that are detected during that 22nd interval. 65 00:04:33,280 --> 00:04:39,040 Now the here we can see that the last 22nd interval ends and the next 22nd interval will start from 66 00:04:39,040 --> 00:04:40,900 here and it will end from here. 67 00:04:40,900 --> 00:04:45,250 And these are the license plate that are detected during that 22nd interval. 68 00:04:45,520 --> 00:04:50,620 So now here you can see this is the cumulative JSON file which contains all the list of the license 69 00:04:50,620 --> 00:04:53,690 plates that are detected during that 20s interval. 70 00:04:54,140 --> 00:04:54,800 Okay. 71 00:04:56,510 --> 00:05:03,020 And over here you can see that output like these are the JSON file that are generated after every 22nd 72 00:05:03,020 --> 00:05:03,410 interval. 73 00:05:03,410 --> 00:05:06,170 So let me show you one of these as well. 74 00:05:09,740 --> 00:05:13,400 So now over here you can see that this is our start time. 75 00:05:13,400 --> 00:05:14,330 This is the end time. 76 00:05:14,330 --> 00:05:18,810 And you can see this is a list of split data directory during that 22nd interval. 77 00:05:18,810 --> 00:05:24,960 So this is our uh this one file look like that is generated after every 20s. 78 00:05:24,960 --> 00:05:25,710 Terrible. 79 00:05:26,100 --> 00:05:26,640 Okay. 80 00:05:26,670 --> 00:05:29,070 Uh, let me open another one over here. 81 00:05:32,250 --> 00:05:38,040 So now you can see over here, uh, this is another JSON file that is being generated after every 22nd 82 00:05:38,040 --> 00:05:38,430 interval. 83 00:05:38,430 --> 00:05:40,740 And we can see the start time and end time. 84 00:05:40,740 --> 00:05:42,870 So this is how it looks like. 85 00:05:43,230 --> 00:05:48,180 Uh, so this is an overview of how we are going to implement this project. 86 00:05:51,600 --> 00:05:56,250 In the step number one, uh, we will see how we can fine tune the Yolov2 model on the license plate 87 00:05:56,250 --> 00:05:57,000 data set. 88 00:05:57,000 --> 00:06:02,220 In the previous video tutorial hike, I have already explained you how you can fine tune the Yolov2 89 00:06:02,310 --> 00:06:04,290 model on any custom data set. 90 00:06:04,770 --> 00:06:08,980 So now I will not go any details over here, I will just give you a general overview. 91 00:06:09,010 --> 00:06:13,480 How you how you can fine tune the YOLO model on the license plate data set. 92 00:06:13,480 --> 00:06:16,750 So over here this is the data set which I will be using. 93 00:06:16,750 --> 00:06:22,600 So if you haven't created account on Roboflow you can simply create an account on Roboflow and download 94 00:06:22,600 --> 00:06:24,280 this data set over here. 95 00:06:25,180 --> 00:06:30,800 And if you just click on this data set over here uh download zip to computer and you can currently use 96 00:06:30,800 --> 00:06:36,200 the Yolo V8 format because we can format is currently not available over here. 97 00:06:36,440 --> 00:06:39,470 Uh, so basically Yolo Britain is built upon the Yolo V8. 98 00:06:39,470 --> 00:06:42,470 So you can use the Yolo V8 format over here. 99 00:06:42,470 --> 00:06:46,220 And you can simply write show download code over here. 100 00:06:46,220 --> 00:06:52,820 And if you just copy this code and add into your Google Colab notebook, you will be able to export 101 00:06:52,820 --> 00:06:56,520 this data set from Roboflow into your Google Colab notebook. 102 00:06:56,640 --> 00:06:59,100 Okay, and this is our how our data set looks like. 103 00:06:59,100 --> 00:07:04,710 We have around 300 images in our data set, and we have only one class which is license plate. 104 00:07:04,830 --> 00:07:05,250 Okay. 105 00:07:05,250 --> 00:07:11,760 And if I just go to the health check over here, like we have 300 images and we have a total of 333 106 00:07:11,760 --> 00:07:13,470 annotations in these 300 images. 107 00:07:13,470 --> 00:07:16,590 And we have only one class, uh, which is license. 108 00:07:16,590 --> 00:07:17,770 And everything looks fine. 109 00:07:17,770 --> 00:07:20,230 Like for one class, 300 images are good enough. 110 00:07:20,230 --> 00:07:23,230 And you can see, uh, this is how our data set look like. 111 00:07:23,230 --> 00:07:28,540 Like you can see we have a bounding box around, uh, license plate, and we have the name written over 112 00:07:28,540 --> 00:07:28,840 here. 113 00:07:28,840 --> 00:07:29,770 License. 114 00:07:29,800 --> 00:07:30,190 Okay. 115 00:07:30,190 --> 00:07:36,610 So we will be using this data set, and we will be fine tuning the model on this license plate data 116 00:07:36,610 --> 00:07:37,180 set. 117 00:07:38,020 --> 00:07:43,340 So first of all, um, I will not run this script again because we need to write down the complete code, 118 00:07:43,460 --> 00:07:44,270 uh, in the next part. 119 00:07:44,270 --> 00:07:46,130 So this is quite a lengthy task. 120 00:07:46,370 --> 00:07:46,760 Okay. 121 00:07:46,760 --> 00:07:49,790 So first of all, you will clone the YOLO and GitHub repo over here. 122 00:07:49,790 --> 00:07:53,150 Then we will you will install all the required packages okay. 123 00:07:53,150 --> 00:07:53,840 Over here. 124 00:07:53,840 --> 00:07:58,010 And then um you will download the pre-trained model weights over here. 125 00:07:58,010 --> 00:08:04,490 And then I've showed you how you can download a data set from Roboflow into your Google Code Colab notebook 126 00:08:04,490 --> 00:08:05,220 over here. 127 00:08:06,420 --> 00:08:12,000 Then we'll fine tune the Yolov2 model on this lesson's Great data set, which will download it from 128 00:08:12,000 --> 00:08:14,640 Roboflow into this Google Colab notebook. 129 00:08:14,640 --> 00:08:20,370 And you can see over here, uh, we have fine tuned the model for 120 epochs. 130 00:08:20,370 --> 00:08:24,900 And let's see what our final mean average precision scores we get okay. 131 00:08:24,900 --> 00:08:31,720 So we have got a mean average precision score with IOU 0.5 as 0.867 and mean average precision score 132 00:08:31,720 --> 00:08:40,060 when IOU varies from 0.5 to 0.95 is 0.486, and we have got a good precision score of 0.90, like 90%, 133 00:08:40,060 --> 00:08:42,430 and the recall score is 79.4%. 134 00:08:42,430 --> 00:08:44,230 So these are good okay. 135 00:08:44,230 --> 00:08:47,080 And this is the like precision confidence curve. 136 00:08:47,080 --> 00:08:51,910 Like uh what we precision score we have got at different confidence thresholds like you can see. 137 00:08:51,910 --> 00:08:53,990 And here we have the recall confidence curve. 138 00:08:54,140 --> 00:08:57,260 Like what recall score we have got at different threshold. 139 00:08:57,260 --> 00:08:59,780 And here we have the precision recall curve. 140 00:08:59,780 --> 00:09:02,630 And uh here we have the F1 confidence curve. 141 00:09:02,810 --> 00:09:07,250 Like what F1 score we have got at different uh confidence threshold like F1 score. 142 00:09:07,280 --> 00:09:09,920 Tell us the trade off between the precision and recall. 143 00:09:10,160 --> 00:09:13,280 Um so basically now we have the confusion matrix. 144 00:09:13,280 --> 00:09:17,290 Confusion matrix tells us basically how our model handles different classes. 145 00:09:17,290 --> 00:09:21,490 So um, here the testing is being performed on the validation data data set. 146 00:09:21,490 --> 00:09:27,430 And now you can see that, uh, around at 56 instances, our model is able to detect correctly that 147 00:09:27,430 --> 00:09:32,830 there is a license plate while and 12 instances when there was a license plate, our model is unable 148 00:09:32,830 --> 00:09:37,720 to detect that, uh, there is a license plate, and with, uh, it is unable to detect anything. 149 00:09:37,720 --> 00:09:38,320 Okay. 150 00:09:40,030 --> 00:09:43,120 So here we have the confusion matrix over here. 151 00:09:43,120 --> 00:09:48,670 Like you can see 82% of the times our model predicted that there is a license plate and 18% of the time 152 00:09:48,670 --> 00:09:50,110 like then there is a license plate. 153 00:09:50,110 --> 00:09:52,030 Our model is unable to detect anything. 154 00:09:52,030 --> 00:09:56,530 And like you can see over here, our mean average precision score is continuously improving. 155 00:09:56,530 --> 00:10:01,690 And uh, as we increase the number of epochs during training and these are the model predictions on 156 00:10:01,690 --> 00:10:02,710 the training batches. 157 00:10:02,710 --> 00:10:05,720 So we can ignore it because we are using the same data for training. 158 00:10:05,720 --> 00:10:08,240 And these are the model prediction on the validation batch. 159 00:10:08,450 --> 00:10:11,870 Uh, we can view it because these images are not used for the training. 160 00:10:11,870 --> 00:10:15,350 So it's always better to take a look and see how our model is performing. 161 00:10:15,350 --> 00:10:19,820 Like you can see there, there we have a number plate, but our model is unable to detect it. 162 00:10:19,820 --> 00:10:23,570 While in all other cases the model is able to detect the license plate. 163 00:10:23,720 --> 00:10:26,750 So after training the model have saved the model weights on the drive. 164 00:10:26,750 --> 00:10:30,450 So I'm directly downloading the model weights from the drive into this Google Colab notebook. 165 00:10:30,450 --> 00:10:33,030 And then I'm doing some inference on these images. 166 00:10:33,030 --> 00:10:36,120 Like you can see over here, we are able to track the license plate. 167 00:10:36,120 --> 00:10:43,350 Then next what I'm doing is that I'm testing my model on a demo video to see if I am able to detect 168 00:10:43,350 --> 00:10:44,760 the license plate or not. 169 00:10:44,760 --> 00:10:48,570 So let me download this video and show you how our output looks like. 170 00:10:53,860 --> 00:10:56,380 So let me just navigate my screen towards it. 171 00:10:59,410 --> 00:11:05,770 So now you can see over here, uh, we are able to detect a license plate like you can see over here. 172 00:11:05,770 --> 00:11:08,050 And the detection results are quite impressive. 173 00:11:08,050 --> 00:11:12,370 Like we are accurately able to detect the license plate by a very good margin. 174 00:11:13,150 --> 00:11:16,510 Okay, so the detection results look quite impressive. 175 00:11:16,570 --> 00:11:21,080 Uh, so now I have shown you how you can fine tune the YOLO model to detect the license plate. 176 00:11:21,080 --> 00:11:23,030 So the step number one is done. 177 00:11:26,120 --> 00:11:30,890 Now, as you can see over here, I have created a new project in my PyCharm community Edition. 178 00:11:30,890 --> 00:11:35,210 So in the step number one we will clone the Yolo v ten GitHub repo over here. 179 00:11:35,210 --> 00:11:40,370 So uh, as you can see over here, this is the official YOLO and GitHub repo. 180 00:11:40,370 --> 00:11:44,280 So I will just pop over here and copy this stuff from here. 181 00:11:45,720 --> 00:11:50,280 And just I will just write it down and I will just add this thing over here. 182 00:11:50,280 --> 00:11:53,970 So now you can see the repository is being cloned over here. 183 00:11:53,970 --> 00:11:57,150 So it will take few more seconds before it gets completed. 184 00:11:57,300 --> 00:11:57,660 Okay. 185 00:11:57,660 --> 00:12:00,390 So now in the next step we will install all the required packages. 186 00:12:00,390 --> 00:12:05,250 So I will just write install hyphen a dot. 187 00:12:05,980 --> 00:12:08,230 So now all the packages will get installed. 188 00:12:08,230 --> 00:12:10,270 So this will take few more seconds. 189 00:12:10,840 --> 00:12:11,560 Okay. 190 00:12:12,280 --> 00:12:12,520 Okay. 191 00:12:12,520 --> 00:12:12,760 No. 192 00:12:12,760 --> 00:12:13,990 Also, I have made a mistake. 193 00:12:13,990 --> 00:12:14,920 I have not set. 194 00:12:15,280 --> 00:12:19,960 Uh, so if I just go over here. 195 00:12:23,770 --> 00:12:24,310 Yeah. 196 00:12:24,520 --> 00:12:27,280 Uh, the repository is being cloned. 197 00:12:33,140 --> 00:12:35,210 Here you can see we have the cloned repository. 198 00:12:35,210 --> 00:12:38,750 I need to set this repository as my current directory. 199 00:12:38,750 --> 00:12:39,290 And. 200 00:12:39,290 --> 00:12:40,040 That's right. 201 00:12:41,180 --> 00:12:41,840 So let's see. 202 00:12:41,840 --> 00:12:43,370 Hopefully this will work. 203 00:12:44,720 --> 00:12:49,250 So now you can see it's installing build dependencies over here. 204 00:12:49,550 --> 00:12:52,820 Uh so this will take some time uh, to get completed. 205 00:12:58,500 --> 00:13:01,530 But in the meanwhile, I will just create a script over here. 206 00:13:09,630 --> 00:13:15,930 And pr.py and pz dot stands for automatic number plate recognition. 207 00:13:16,590 --> 00:13:17,250 Okay. 208 00:13:18,130 --> 00:13:21,670 So this now all the packages are being getting installed. 209 00:13:21,670 --> 00:13:24,640 So this will take some time like you can see over here. 210 00:13:26,260 --> 00:13:28,090 So this is a bit time consuming. 211 00:13:43,970 --> 00:13:46,220 So we will also okay. 212 00:13:46,220 --> 00:13:48,350 So this package is also installed. 213 00:13:48,350 --> 00:13:50,480 So we don't need to install it separately. 214 00:13:50,780 --> 00:13:55,610 Okay I think um all the packages are being getting installed over here. 215 00:13:58,730 --> 00:14:04,130 So now I will just create another folder into this project with the name resources, where I will add 216 00:14:04,130 --> 00:14:10,590 some images and videos so that I can test, uh, do the testing later on, uh, of our model, like 217 00:14:10,590 --> 00:14:14,280 whether it is able to detect the license plate or read the license plate or not. 218 00:14:16,770 --> 00:14:21,930 Here you can see that I have just created a folder inside this project over here, uh, by the name 219 00:14:21,930 --> 00:14:22,530 resources. 220 00:14:22,530 --> 00:14:24,240 Here you can see we have some car images. 221 00:14:24,240 --> 00:14:27,960 And here we have some videos of cars over here. 222 00:14:27,960 --> 00:14:33,670 So we will be detecting and recognizing the license plate in this video as well as in these images as 223 00:14:33,670 --> 00:14:34,090 well. 224 00:14:34,090 --> 00:14:37,540 So let's go back to the project over here okay. 225 00:14:37,540 --> 00:14:44,050 So now if I just refresh this over here, we can see the uh resources folder over here. 226 00:14:44,260 --> 00:14:44,440 Yeah. 227 00:14:45,940 --> 00:14:46,750 So. 228 00:14:50,770 --> 00:14:52,000 So let's get started. 229 00:14:52,000 --> 00:15:00,320 So first of all, I will just write import cv2 and from. 230 00:15:00,590 --> 00:15:03,830 So let's uh only read the video at the start. 231 00:15:05,450 --> 00:15:08,660 So first we will create a video capture object over here. 232 00:15:13,400 --> 00:15:14,000 Capture. 233 00:15:14,810 --> 00:15:19,050 And we will go inside the resources folder. 234 00:15:22,410 --> 00:15:32,250 And from there we will just write our lessons one dot mp4. 235 00:15:35,880 --> 00:15:36,480 Okay. 236 00:15:43,690 --> 00:15:46,000 We can add comments over here as well. 237 00:15:46,000 --> 00:15:49,720 Uh, first we will create a, uh, import all the required libraries. 238 00:15:53,650 --> 00:15:56,770 All the required libraries. 239 00:15:58,330 --> 00:16:02,830 And here we are creating our video capture object. 240 00:16:09,640 --> 00:16:10,420 Can you not? 241 00:16:19,420 --> 00:16:21,460 Read this. 242 00:16:22,120 --> 00:16:23,110 If read. 243 00:16:25,900 --> 00:16:30,710 CV2 dot I am show radio. 244 00:16:32,750 --> 00:16:38,420 Name and if CV two dot wait key one. 245 00:16:53,850 --> 00:16:55,140 So then it will break. 246 00:17:14,790 --> 00:17:20,500 And here we can simply write cap dot release. 247 00:17:22,780 --> 00:17:26,380 And we can also write cv2 dot destroy all windows. 248 00:17:34,120 --> 00:17:37,150 In this file. 249 00:17:45,110 --> 00:17:47,300 Okay, so let's run this up now. 250 00:17:48,110 --> 00:17:50,630 I think the indexing will take some time to collate. 251 00:17:50,630 --> 00:17:52,460 Then we will be able to run this up. 252 00:17:52,970 --> 00:17:56,690 So let's wait for the indexing to get complete and then we'll run the script. 253 00:18:00,500 --> 00:18:02,300 Now you can see the indexing is being done. 254 00:18:02,300 --> 00:18:04,670 So let's run this script and let's see. 255 00:18:05,060 --> 00:18:08,070 Now you can see that, uh, the video is running very fine. 256 00:18:08,070 --> 00:18:14,130 So if I just want to run another video, let's right over here and run r.com. 257 00:18:15,000 --> 00:18:20,280 Now you can see the video is running very smoothly and a good frame rate as well. 258 00:18:20,460 --> 00:18:23,280 And you can see that, uh the frame rate is quite good. 259 00:18:23,280 --> 00:18:23,850 Yeah. 260 00:18:24,960 --> 00:18:25,500 Okay. 261 00:18:25,650 --> 00:18:30,610 So everything works currently looking very smooth and fine. 262 00:18:30,850 --> 00:18:37,960 Um, one thing, uh, we can do, but, uh, next, like, uh, let's add the frame rate over here, 263 00:18:38,320 --> 00:18:40,660 or let's keep it away from now. 264 00:18:40,720 --> 00:18:45,070 Uh, we can, uh, do it further on for now. 265 00:18:45,070 --> 00:18:50,020 First, uh, now we will see how we can do, uh, detect the license plate and read the text from the 266 00:18:50,020 --> 00:18:51,010 latent license plate. 267 00:18:51,010 --> 00:18:54,020 So, um, I will just write from ultralytics. 268 00:18:54,620 --> 00:18:57,740 Uh, import yolo v ten. 269 00:18:58,310 --> 00:18:58,730 Okay. 270 00:18:58,730 --> 00:19:00,050 That's good. 271 00:19:03,500 --> 00:19:06,470 Then I will just go over here and just write. 272 00:19:08,630 --> 00:19:20,460 I will just initialize the yolo v and model The site model is equal to yolo v ten. 273 00:19:21,660 --> 00:19:29,640 And then I will just go inside this uh weights directory where I have added the model best weights over 274 00:19:29,640 --> 00:19:30,150 here. 275 00:19:30,540 --> 00:19:33,210 So I will just go inside the weights factory and the site. 276 00:19:33,210 --> 00:19:34,230 best.bt. 277 00:19:35,040 --> 00:19:35,700 Okay. 278 00:19:38,370 --> 00:19:40,080 So that looks good. 279 00:19:45,430 --> 00:19:49,570 Plus, we can initialize the frame count over here as well. 280 00:19:49,600 --> 00:19:54,880 Initialize the frame count and we'll say count is equal to zero. 281 00:19:54,880 --> 00:20:00,160 And here we can write count plus one. 282 00:20:00,610 --> 00:20:04,390 Uh print frame count. 283 00:20:09,380 --> 00:20:09,740 City. 284 00:20:09,740 --> 00:20:10,070 Right. 285 00:20:11,690 --> 00:20:11,960 Name. 286 00:20:11,960 --> 00:20:12,770 Number. 287 00:20:15,380 --> 00:20:17,300 Okay, so let's look promising. 288 00:20:20,150 --> 00:20:25,550 And let's write a result model dot date. 289 00:20:25,550 --> 00:20:25,560 day. 290 00:20:29,910 --> 00:20:33,900 And we can set the conference table to 0.45 over here. 291 00:20:37,710 --> 00:20:41,400 And we can write for result in results. 292 00:20:41,400 --> 00:20:44,490 So we'll loop through uh, each of the bounding boxes. 293 00:20:44,490 --> 00:20:48,510 And uh, then we will draw bounding boxes on each of the license plates. 294 00:20:48,510 --> 00:20:57,670 So first we'll look through each of the bounding boxes and simply write boxes equal to result dot boxes. 295 00:20:59,560 --> 00:20:59,980 Right. 296 00:20:59,980 --> 00:21:03,010 For box boxes. 297 00:21:05,260 --> 00:21:08,710 We have x1 y1 x2 y2. 298 00:21:09,400 --> 00:21:09,790 Okay. 299 00:21:09,790 --> 00:21:13,150 And we can write box dot x y x y. 300 00:21:13,150 --> 00:21:13,160 Why? 301 00:21:17,000 --> 00:21:20,660 So the output which you get from here is in the form of tensors. 302 00:21:20,840 --> 00:21:23,270 Well, let me show you as well. 303 00:21:34,370 --> 00:21:37,940 So let's see if it works fine or if there is an error then we'll fix this up. 304 00:21:44,190 --> 00:21:47,820 Okay, so this will take some time before it gets started to run. 305 00:21:48,300 --> 00:21:48,690 Okay. 306 00:21:48,690 --> 00:21:49,620 We have an error. 307 00:21:49,620 --> 00:21:51,120 So let's see what's that error. 308 00:21:56,520 --> 00:22:00,480 Also we are just facing an error here. 309 00:22:01,290 --> 00:22:01,650 Okay. 310 00:22:01,650 --> 00:22:05,980 So the issue we are just getting is uh this is not installed. 311 00:22:06,490 --> 00:22:06,880 Okay. 312 00:22:06,880 --> 00:22:09,970 So we can simply write, uh. 313 00:22:13,900 --> 00:22:16,030 And let's check out this error. 314 00:22:16,360 --> 00:22:21,520 We don't know what exact package I installed. 315 00:22:21,790 --> 00:22:23,140 Uh, what is this? 316 00:22:24,220 --> 00:22:24,490 Okay. 317 00:22:24,490 --> 00:22:29,080 So we can simply copy this from here and just like. 318 00:22:29,450 --> 00:22:31,880 so, let's see if this works or not. 319 00:22:41,510 --> 00:22:41,720 Okay. 320 00:22:41,720 --> 00:22:43,220 So let's run this again. 321 00:22:48,770 --> 00:22:49,070 Okay. 322 00:22:49,070 --> 00:22:50,900 So it's currently updating indexes. 323 00:22:50,900 --> 00:22:52,220 So this will take some time. 324 00:22:52,220 --> 00:22:55,290 So let's wait for it to get complete. 325 00:22:59,940 --> 00:23:00,870 I think that's done. 326 00:23:18,190 --> 00:23:19,630 And what is there? 327 00:23:23,590 --> 00:23:26,140 NumPy is not available. 328 00:23:35,920 --> 00:23:36,160 Well. 329 00:23:41,870 --> 00:23:44,870 I just found out the solution of the problem that we are facing. 330 00:23:45,050 --> 00:23:50,000 First we need to uninstall numpy and then we need to install numpy with this version. 331 00:23:50,000 --> 00:23:52,190 So this will hopefully solve our problem. 332 00:23:52,190 --> 00:23:53,900 So let's do this. 333 00:23:55,610 --> 00:23:58,340 The meanwhile it gets uninstalled. 334 00:23:58,340 --> 00:24:00,110 Let's run this up now. 335 00:24:05,150 --> 00:24:09,870 Okay so now we are installing this version and hopefully this will work. 336 00:24:10,620 --> 00:24:11,490 Okay. 337 00:24:21,780 --> 00:24:23,310 Currently we are over here. 338 00:24:30,400 --> 00:24:36,490 So now it's just completing the indexing and it will take few more seconds before it gets completed. 339 00:24:47,530 --> 00:24:48,490 Um, next. 340 00:24:48,490 --> 00:24:50,080 Okay, so it's been done. 341 00:24:50,080 --> 00:24:51,430 So let's run this up. 342 00:24:57,380 --> 00:24:59,090 Let's see how does it works. 343 00:25:04,070 --> 00:25:04,370 Okay. 344 00:25:04,370 --> 00:25:06,140 So we have the frame number one over here. 345 00:25:06,140 --> 00:25:07,280 But currently okay. 346 00:25:07,280 --> 00:25:08,750 So here we have the video. 347 00:25:08,780 --> 00:25:10,310 Now currently no detection will be done. 348 00:25:10,310 --> 00:25:13,910 But you can see over here we have the bounding box coordinates over here. 349 00:25:14,120 --> 00:25:15,830 Uh currently it's in the form of tensors. 350 00:25:15,830 --> 00:25:18,000 We need to convert it into integers. 351 00:25:18,030 --> 00:25:20,490 Okay, so if I just press the one key this will stop. 352 00:25:20,490 --> 00:25:21,930 So let's write the rest of the code. 353 00:25:22,530 --> 00:25:24,330 Uh, we can simply move this up. 354 00:25:49,000 --> 00:25:50,620 Then we have the frame. 355 00:25:50,740 --> 00:25:54,730 Let's draw the bounding boxes around the license plates. 356 00:26:00,010 --> 00:26:02,410 The color of the bounding box will be below. 357 00:26:05,140 --> 00:26:06,320 And thickness will be two. 358 00:26:07,580 --> 00:26:08,330 Okay. 359 00:26:08,330 --> 00:26:11,510 And what different classes we have over here. 360 00:26:16,550 --> 00:26:22,460 So we have only one class uh, which is uh, license. 361 00:26:24,440 --> 00:26:24,890 Okay. 362 00:26:29,580 --> 00:26:34,950 So the YOLO written model will give us the class name in the format integers. 363 00:26:36,060 --> 00:26:36,900 So we have. 364 00:26:36,930 --> 00:26:43,110 We can get it by writing box dots plus zero okay. 365 00:26:56,610 --> 00:27:02,340 And now we'll also calculate the confidence score for this math. 366 00:27:18,700 --> 00:27:19,120 Okay. 367 00:27:22,600 --> 00:27:24,490 Then we have the label as. 368 00:27:30,550 --> 00:27:31,390 Last name. 369 00:27:32,500 --> 00:27:33,490 Confidence. 370 00:27:43,250 --> 00:27:45,080 Uh, let's get the text size. 371 00:27:45,080 --> 00:27:48,080 So we'll just write cv2 dot get text size. 372 00:27:52,760 --> 00:27:55,010 One scale will be 0.5. 373 00:27:56,810 --> 00:27:58,430 Thickness of the text will be two. 374 00:28:02,120 --> 00:28:07,230 So we will display the uh uh label which contain the class name and confidence for above the bounding 375 00:28:07,230 --> 00:28:07,800 box. 376 00:28:08,070 --> 00:28:11,220 So for this we can simply write text size zero. 377 00:28:26,100 --> 00:28:28,320 Then I will start with tango. 378 00:28:28,350 --> 00:28:29,740 Over here we have the frame. 379 00:28:34,540 --> 00:28:35,050 Due to. 380 00:28:37,960 --> 00:28:43,420 The color of the bounding box below, which will be added above the bounding detected license plate. 381 00:28:43,420 --> 00:28:46,450 And it will be filled with the blue color. 382 00:28:49,000 --> 00:28:52,000 And we will just put text inside this. 383 00:29:03,950 --> 00:29:06,770 The color of the text will be white. 384 00:29:21,180 --> 00:29:23,850 Okay, so everything looks perfect for now. 385 00:29:27,060 --> 00:29:28,320 Let's run this up now. 386 00:29:44,560 --> 00:29:47,350 So now you can see over here we are able to detect the license plate. 387 00:29:47,350 --> 00:29:49,300 And you can see we have drawn a bounding box. 388 00:29:49,630 --> 00:29:51,670 Uh here we have the class name license. 389 00:29:51,670 --> 00:29:54,220 And here you can see we have the confidence score over here. 390 00:29:54,640 --> 00:29:56,320 Um, the results looks very promising. 391 00:29:56,320 --> 00:29:57,940 Like, the results are good now. 392 00:29:58,840 --> 00:30:01,180 Okay, so let's go ahead. 393 00:30:01,720 --> 00:30:03,610 Let's read the text from license plate. 394 00:30:04,690 --> 00:30:07,190 So we will just write. 395 00:30:07,400 --> 00:30:10,130 We need to install the OCR package. 396 00:30:13,130 --> 00:30:15,500 So let's see how we can do it. 397 00:30:16,220 --> 00:30:22,010 So we need to just install paddle OCR on here so we can just write. 398 00:30:25,040 --> 00:30:26,180 Install paddle OCR. 399 00:30:43,560 --> 00:30:45,240 Uh, meanwhile, the package gets installed. 400 00:30:45,240 --> 00:30:47,520 We can start writing down the code as well. 401 00:30:50,160 --> 00:30:52,410 From OCR import. 402 00:30:55,600 --> 00:30:56,440 Final OCR. 403 00:31:05,110 --> 00:31:05,440 Only. 404 00:31:05,470 --> 00:31:08,650 Here we will initialize the OCR. 405 00:31:12,610 --> 00:31:13,180 OCR. 406 00:31:13,180 --> 00:31:13,250 are. 407 00:31:23,090 --> 00:31:26,990 I don't have a GPU in my local system, so I will just set it to false. 408 00:31:32,720 --> 00:31:35,660 So now we'll also create a function over here for this. 409 00:31:42,090 --> 00:31:44,130 In the frame and the bounding box. 410 00:31:44,130 --> 00:31:44,880 Coordinates. 411 00:32:08,860 --> 00:32:09,220 Okay. 412 00:32:11,320 --> 00:32:14,080 Uh, we don't want to use the pedals here for reduction. 413 00:32:14,080 --> 00:32:16,540 We only want to do recognition using pedal OCR. 414 00:32:30,830 --> 00:32:33,380 And we only want to read that text from the sense plate. 415 00:32:38,420 --> 00:32:41,480 So if nothing is text is read, then this will be empty. 416 00:32:43,100 --> 00:32:49,880 But then we will loop through all the lessons played and we will just read the lesson. 417 00:32:49,880 --> 00:32:51,680 Split text from lessons played. 418 00:32:51,680 --> 00:32:51,690 plate. 419 00:33:00,360 --> 00:33:03,420 Well, I have already done the necessary calculation like this. 420 00:33:03,420 --> 00:33:05,040 Index contain the conference code. 421 00:33:05,040 --> 00:33:09,930 So whenever we read the text from the license plate, we get the confidence code, like how much our 422 00:33:09,930 --> 00:33:17,190 OCR pedal OCR model is confident that, uh, this, uh, that uh, text that the model model model has 423 00:33:17,190 --> 00:33:18,580 read, uh, please. 424 00:33:18,580 --> 00:33:22,570 Like, this is 60% confident that this is the correct text that it has read. 425 00:33:23,320 --> 00:33:25,480 So here we have the following text. 426 00:33:25,480 --> 00:33:26,200 We have the score. 427 00:33:26,200 --> 00:33:32,320 If you want to check more details you can simply write print OCR and just print out the text that it 428 00:33:32,320 --> 00:33:32,710 has read. 429 00:33:32,710 --> 00:33:34,030 So you can have a better idea. 430 00:33:34,030 --> 00:33:37,870 But I'm not doing it correctly because this is already taking so much time. 431 00:33:47,450 --> 00:33:52,370 So if the score is not available, then you set the score is equal to zero. 432 00:33:53,690 --> 00:33:55,820 Else if the score is available. 433 00:33:58,700 --> 00:34:04,370 Then we'll simply multiply the score by 100, because the score will be in the form of 0.60.1 or so. 434 00:34:07,350 --> 00:34:17,790 And if force is greater than 60%, like if the model is 60% confident or more, then we will just write 435 00:34:17,790 --> 00:34:19,560 x or zero. 436 00:34:32,350 --> 00:34:34,330 So after we have written the text. 437 00:34:34,330 --> 00:34:39,970 So now we need to import free so that we can remove any irregularities in the text. 438 00:35:22,890 --> 00:35:26,310 So if there are triple question marks or like such like this stuff. 439 00:35:26,310 --> 00:35:29,580 So we will replace this in our text as well. 440 00:35:29,700 --> 00:35:36,240 And if the text we have the O we will replace the O with zero. 441 00:35:48,280 --> 00:35:50,260 Similarly if we have this as well. 442 00:35:53,800 --> 00:35:55,780 So we have created a function over here. 443 00:35:55,780 --> 00:35:58,210 So let's write down over here as well. 444 00:36:02,380 --> 00:36:04,570 So we can comment this out correctly. 445 00:36:04,570 --> 00:36:06,710 And we can simply write label. 446 00:36:08,900 --> 00:36:11,660 Is equal to pad. 447 00:36:13,940 --> 00:36:14,570 OCR. 448 00:36:14,570 --> 00:36:17,240 And here we have name. 449 00:36:22,490 --> 00:36:23,060 That's good. 450 00:36:26,720 --> 00:36:29,480 So let's run this up and see how does it works. 451 00:36:50,370 --> 00:36:50,640 For. 452 00:36:56,770 --> 00:36:57,400 Yeah, okay. 453 00:37:00,370 --> 00:37:02,860 So I've installed already installed Docker. 454 00:37:02,860 --> 00:37:03,130 But. 455 00:38:03,710 --> 00:38:04,310 So it's done. 456 00:38:04,310 --> 00:38:05,720 So let's run this up now. 457 00:38:09,360 --> 00:38:09,570 Okay. 458 00:38:09,570 --> 00:38:11,790 So the indexing will take some time to complete. 459 00:38:21,960 --> 00:38:24,180 So let's wait for it to get completed. 460 00:38:24,180 --> 00:38:25,710 Then we can go ahead. 461 00:38:39,310 --> 00:38:41,140 No updating Skelton. 462 00:38:41,140 --> 00:38:42,790 So this will not take very much time. 463 00:38:54,800 --> 00:38:56,990 And installed packages switched on. 464 00:38:59,330 --> 00:39:00,680 So let's run this up now. 465 00:39:26,100 --> 00:39:29,250 Okay, I'm already open. 466 00:39:29,370 --> 00:39:29,640 Okay. 467 00:39:29,640 --> 00:39:32,100 So let's resolve this error as well. 468 00:39:32,100 --> 00:39:34,740 So we'll just write port OS. 469 00:39:46,660 --> 00:39:47,320 AMP. 470 00:39:47,320 --> 00:39:48,040 Dash. 471 00:39:49,390 --> 00:39:50,380 Duplicate. 472 00:40:02,500 --> 00:40:04,270 I hope this will solve this error. 473 00:40:07,370 --> 00:40:11,180 So that we can solve any, uh, duplicate library issues. 474 00:40:23,870 --> 00:40:25,490 Let's see how does it works. 475 00:40:31,230 --> 00:40:37,320 So now you can see over here we are able to read the text from the license plate like our 183JF. 476 00:40:37,950 --> 00:40:41,790 So we are able to read the text from the license plate and it looks good. 477 00:40:42,420 --> 00:40:47,220 So let's see if we are able to read the text from the other license plates as well or not. 478 00:40:47,370 --> 00:40:47,550 Um. 479 00:40:49,560 --> 00:40:50,040 Let's see. 480 00:40:51,330 --> 00:40:51,570 Okay. 481 00:40:52,380 --> 00:40:55,930 For this current license plate, we are able to read the text from the license plate. 482 00:40:55,930 --> 00:41:00,160 So now you can see that we have integrated YOLO v ten and OCR code. 483 00:41:00,160 --> 00:41:05,230 We are using Yolo v ten to detect the license plate and to read the text from the license plate. 484 00:41:05,230 --> 00:41:07,360 We are using paddle OCR. 485 00:41:08,050 --> 00:41:08,350 Okay. 486 00:41:08,350 --> 00:41:13,450 So currently you can see the text is not very clear, like it's N8N9 for j v. 487 00:41:13,450 --> 00:41:16,570 So we are able to read the text on the license plate here as well. 488 00:41:16,570 --> 00:41:18,290 So that looks amazing. 489 00:41:18,290 --> 00:41:23,720 Like we have used pedal ACR to read the text from the license plate and the results are very impressive. 490 00:41:23,930 --> 00:41:29,870 Okay, so now the next step we need to, uh, save the license plate data into a JSON file. 491 00:41:29,870 --> 00:41:31,670 So let's work towards it. 492 00:41:35,060 --> 00:41:40,640 So I will just create a function over here save JSON okay. 493 00:41:40,640 --> 00:41:43,560 In the input will be the unique license plates. 494 00:41:43,920 --> 00:41:44,520 Okay. 495 00:41:47,130 --> 00:41:56,100 And the start time, because we need, uh, to save the license plates, uh, data after every 20s interval. 496 00:42:00,630 --> 00:42:12,970 So we need to generate individual JSON files for each 22nd interval. 497 00:42:13,390 --> 00:42:14,110 Okay. 498 00:42:15,430 --> 00:42:20,020 So in the interval data we have. 499 00:42:25,060 --> 00:42:27,370 We have our time. 500 00:42:31,430 --> 00:42:31,700 Okay. 501 00:42:34,220 --> 00:42:36,260 Start time will be in the ISO format. 502 00:42:38,540 --> 00:42:40,640 Similarly, we also have the end time. 503 00:42:43,400 --> 00:42:46,610 And end time will be also in the ISO format. 504 00:42:49,220 --> 00:42:50,840 Then we have the license plate. 505 00:42:50,840 --> 00:42:53,930 And the license plate will be in the form of lists. 506 00:43:05,070 --> 00:43:09,390 And the interval file path will be as following. 507 00:43:13,200 --> 00:43:18,670 Also we need to create a directory over here by the name Jason. 508 00:43:19,660 --> 00:43:23,440 And inside this will write a data file with the name output. 509 00:43:36,610 --> 00:43:39,670 And here we have um air. 510 00:43:39,670 --> 00:43:39,680 Year. 511 00:43:42,590 --> 00:43:43,400 Month. 512 00:43:43,910 --> 00:43:44,420 Day. 513 00:43:45,290 --> 00:43:46,310 Hours. 514 00:43:46,850 --> 00:43:47,720 Minutes. 515 00:43:47,720 --> 00:43:48,590 Seconds. 516 00:43:50,810 --> 00:43:54,410 And just save it in the form of a JSON. 517 00:43:54,590 --> 00:43:55,970 Okay, so that's no good. 518 00:43:58,010 --> 00:43:59,930 And we'll just right click open. 519 00:44:10,590 --> 00:44:11,700 Interval file path. 520 00:44:22,110 --> 00:44:23,100 Interval data. 521 00:44:23,220 --> 00:44:27,840 And inside this, uh, JSON file, we build up this interval data. 522 00:44:36,250 --> 00:44:38,080 And we have the intern test. 523 00:44:38,110 --> 00:44:39,430 So that's good. 524 00:44:39,880 --> 00:44:44,380 So this will generate individual generation file for each 22nd interval. 525 00:44:45,970 --> 00:44:49,300 Then next we need to generate a cumulative JSON file as well. 526 00:44:49,300 --> 00:44:50,080 So. 527 00:44:59,150 --> 00:45:01,220 we need to define the cumulative. 528 00:45:05,450 --> 00:45:07,070 File path as well. 529 00:45:09,020 --> 00:45:11,390 And this will be inside the JSON directory. 530 00:45:11,390 --> 00:45:13,400 And the file name will be less since. 531 00:45:15,410 --> 00:45:17,960 The data project. 532 00:45:20,700 --> 00:45:21,990 And then we'll just try to. 533 00:45:21,990 --> 00:45:26,670 If there exists a cumulative file path. 534 00:45:31,890 --> 00:45:34,050 Then we'll open the cumulative file. 535 00:45:44,920 --> 00:45:45,490 Read it. 536 00:45:52,840 --> 00:45:55,630 Then we'll load the existing data over here. 537 00:45:58,690 --> 00:46:00,790 That we have inside this file. 538 00:46:03,850 --> 00:46:09,410 And the existing Existing data will be empty. 539 00:46:15,980 --> 00:46:20,810 So now we'll add new interval data to this cumulative file. 540 00:46:37,890 --> 00:46:38,940 Existing data dot. 541 00:46:38,940 --> 00:46:39,840 Append. 542 00:46:42,300 --> 00:46:43,110 One data. 543 00:46:45,810 --> 00:46:47,700 I can simply write will open. 544 00:47:13,030 --> 00:47:14,260 So that looks good. 545 00:47:21,020 --> 00:47:23,630 So now next week, what we'll do is. 546 00:47:34,220 --> 00:47:36,170 So now we can just try it over here. 547 00:47:39,650 --> 00:47:41,030 So we have. 548 00:47:43,820 --> 00:47:44,810 All this stuff. 549 00:47:46,970 --> 00:47:50,720 Where I have initialized, we have. 550 00:47:50,720 --> 00:47:52,490 Start time is equal to. 551 00:47:54,890 --> 00:47:55,220 Date. 552 00:47:55,220 --> 00:47:55,520 Time. 553 00:47:55,520 --> 00:47:55,820 Dot. 554 00:47:55,820 --> 00:47:56,330 Now. 555 00:48:02,600 --> 00:48:04,130 Here we have the current timers. 556 00:48:12,570 --> 00:48:14,040 Okay, so that looks good. 557 00:48:23,040 --> 00:48:24,870 Next what we will do over here is. 558 00:48:31,540 --> 00:48:35,320 Now we'll just separate the 20s time interval. 559 00:48:47,140 --> 00:48:54,770 So if we have the time is greater or equal to 20s, then we have the end time. 560 00:48:56,600 --> 00:48:56,930 Run. 561 00:48:59,930 --> 00:49:00,470 Time. 562 00:49:05,270 --> 00:49:11,150 Okay, so first we need to save the license plate data like we have to accumulate. 563 00:49:11,300 --> 00:49:16,370 Add unique license plates into our JSON file okay. 564 00:49:16,370 --> 00:49:16,380 Okay. 565 00:49:18,390 --> 00:49:20,640 So for this I will just write over here. 566 00:49:28,560 --> 00:49:29,160 Okay. 567 00:49:32,700 --> 00:49:37,380 So if we have read some text then we'll just write if label. 568 00:49:43,600 --> 00:49:47,770 So if we have label, then we will add this unique license plate over here. 569 00:49:49,120 --> 00:49:49,690 Okay. 570 00:49:51,940 --> 00:49:55,630 Then I will just write over here license plates. 571 00:49:57,550 --> 00:50:03,280 Then we have the start time and time over here. 572 00:50:04,090 --> 00:50:04,480 Okay. 573 00:50:07,610 --> 00:50:08,000 Okay. 574 00:50:09,710 --> 00:50:13,310 Then we have the start time as run time. 575 00:50:13,310 --> 00:50:14,090 Over here. 576 00:50:14,960 --> 00:50:15,500 Okay. 577 00:50:15,500 --> 00:50:15,980 Start time. 578 00:50:16,610 --> 00:50:17,660 That's fine. 579 00:50:22,070 --> 00:50:25,130 And then we will just clear all this data. 580 00:50:28,640 --> 00:50:31,860 For the next 22nd, Jake's on file. 581 00:50:32,250 --> 00:50:35,520 So let's run this and see how it works. 582 00:50:56,230 --> 00:50:58,210 So this will take few more seconds. 583 00:51:01,300 --> 00:51:01,630 Okay. 584 00:51:01,630 --> 00:51:03,040 Where has this error? 585 00:51:19,910 --> 00:51:23,150 Okay, so we need to write from date time. 586 00:51:24,290 --> 00:51:24,800 Okay. 587 00:51:25,040 --> 00:51:26,810 I think this will solve the error. 588 00:51:26,810 --> 00:51:27,260 Let's see. 589 00:51:49,920 --> 00:51:53,250 So now you can see that we are able to create the license plate. 590 00:51:53,730 --> 00:51:53,940 Okay. 591 00:51:53,940 --> 00:52:00,690 So let me open this up as well and see if we are able to save the status. 592 00:52:05,850 --> 00:52:07,990 So here we will have a JSON files. 593 00:52:07,990 --> 00:52:10,060 But let's see what does it do. 594 00:52:10,600 --> 00:52:11,800 I think there's an error. 595 00:52:13,270 --> 00:52:13,810 Okay. 596 00:52:13,810 --> 00:52:15,670 So let's fix up this error. 597 00:52:16,000 --> 00:52:16,180 Uh. 598 00:52:32,690 --> 00:52:33,440 Okay. 599 00:52:37,880 --> 00:52:39,350 I think this will fix this up. 600 00:52:39,350 --> 00:52:40,490 And let's run this again. 601 00:53:06,120 --> 00:53:09,510 So now you can see we are able to read the text of license plates over here. 602 00:53:12,210 --> 00:53:15,450 And let's see if we have the JSON file over here or not. 603 00:53:25,240 --> 00:53:26,530 Currently it's empty. 604 00:53:27,280 --> 00:53:28,150 Okay. 605 00:53:37,210 --> 00:53:37,720 Okay. 606 00:53:37,720 --> 00:53:39,940 I just got this error over here. 607 00:53:39,940 --> 00:53:43,550 Let's run this again now and see how does it works. 608 00:53:47,900 --> 00:53:50,780 Now let's run this again and see if it works or not. 609 00:53:50,780 --> 00:53:53,180 So I will just run this script now. 610 00:54:04,250 --> 00:54:08,940 So it should generate a JSON file after every 22nd interval. 611 00:54:08,940 --> 00:54:14,250 So now you can see over here, uh, it will generate individual JSON files for each 22nd interval. 612 00:54:14,250 --> 00:54:20,280 And there will be a cumulative JSON file, uh, which will contain the timestamps for all the intervals 613 00:54:20,280 --> 00:54:24,060 as well as the license plate data for all the intervals as well. 614 00:54:24,060 --> 00:54:29,070 So now it has started and we will see our JSON files in in this folder. 615 00:54:29,070 --> 00:54:30,240 Currently it's empty. 616 00:54:30,340 --> 00:54:34,390 Uh, it will be generating JSON files after every 22nd interval. 617 00:54:34,390 --> 00:54:36,160 So let's see. 618 00:54:37,210 --> 00:54:37,540 Okay. 619 00:54:37,540 --> 00:54:41,140 So now we have our first, uh, JSON file over here. 620 00:54:41,140 --> 00:54:41,380 Okay. 621 00:54:41,380 --> 00:54:43,000 So let's uh see. 622 00:54:43,000 --> 00:54:49,330 So now you can see over here we have uh our interval like 4440 and 45. 623 00:54:49,330 --> 00:54:53,890 And here you can see this is the state that have been detected during this 22nd interval. 624 00:54:53,890 --> 00:54:56,450 And this is the cumulative JSON file over here. 625 00:54:56,450 --> 00:55:02,570 And you can see over here the same interval and the data which has been detected. 626 00:55:02,570 --> 00:55:09,200 So now we can see after a 22nd interval we have another JSON file which is from this interval to this 627 00:55:09,200 --> 00:55:09,680 interval. 628 00:55:09,680 --> 00:55:13,370 And this is the license plate data detected during this interval. 629 00:55:14,420 --> 00:55:15,680 And this is a completed file. 630 00:55:15,680 --> 00:55:20,700 So now you can see this is the uh license plate detected during this first 22nd interval. 631 00:55:20,700 --> 00:55:24,630 And these are the license plates that are detected during the next 22nd interval. 632 00:55:24,900 --> 00:55:25,410 Okay. 633 00:55:26,700 --> 00:55:28,890 So now this is the third JSON file. 634 00:55:28,890 --> 00:55:33,420 So now you can see over here these are the license plate data detected during this interval provided 635 00:55:33,420 --> 00:55:34,170 over here. 636 00:55:34,170 --> 00:55:41,610 And this is accumulated this file which contains um all the data data for all the intervals which have 637 00:55:41,610 --> 00:55:42,580 been currently processed. 638 00:55:44,320 --> 00:55:46,360 So now this is how it works. 639 00:55:46,360 --> 00:55:47,830 So I will not stop this code. 640 00:55:47,830 --> 00:55:53,110 Uh uh uh, currently it will start continue producing the output like you can see over here. 641 00:55:53,110 --> 00:55:57,430 I will not stop the script currently, but uh, we will go towards the next part. 642 00:55:58,390 --> 00:56:03,280 So the next we will see how we can save the license plate data to the SQL database. 643 00:56:03,280 --> 00:56:05,590 So now we will go ahead towards that part. 644 00:56:08,000 --> 00:56:09,980 So we will be using SQLite. 645 00:56:10,220 --> 00:56:12,770 Um so which is a very lightweight. 646 00:56:12,860 --> 00:56:14,330 Lightweight disk based database. 647 00:56:14,330 --> 00:56:17,540 And we will be using the SQL SQLite. 648 00:56:19,070 --> 00:56:21,770 And it is very easy to set up and use as well. 649 00:56:21,770 --> 00:56:25,370 So SQLite SQLite comes pre-installed. 650 00:56:25,370 --> 00:56:26,870 So we not need to install it. 651 00:56:26,870 --> 00:56:30,540 But we need to install uh the SQLite three module. 652 00:56:30,540 --> 00:56:32,550 So let's get started with this. 653 00:56:34,920 --> 00:56:35,640 Just right. 654 00:56:39,600 --> 00:56:43,530 So we'll just copy this from here. 655 00:56:44,460 --> 00:56:48,090 We install this so we can simply go to settings. 656 00:56:49,320 --> 00:56:51,660 Um here we have. 657 00:56:56,530 --> 00:56:59,440 Uh, now here we click on the plus sign. 658 00:57:14,710 --> 00:57:16,510 So here we need to install this package. 659 00:57:16,510 --> 00:57:20,420 So you can simply write, uh, this install this package from here. 660 00:57:20,480 --> 00:57:23,960 Or you can simply write pip install in the terminal as well. 661 00:57:23,960 --> 00:57:25,910 So this is how you can do this. 662 00:57:28,910 --> 00:57:35,300 So here you can simply write import Sqlite3 over here. 663 00:57:42,270 --> 00:57:47,340 So first we need to create an SQLite database and a table to store the license plate data. 664 00:57:47,370 --> 00:57:50,910 So first we need to create a SQLite database and table. 665 00:57:50,910 --> 00:57:52,920 So I will just write SQL. 666 00:57:53,790 --> 00:57:53,970 Uh. 667 00:58:00,000 --> 00:58:01,860 What should I write the name over here. 668 00:58:18,790 --> 00:58:19,630 I will just try. 669 00:58:20,470 --> 00:58:26,440 So first we will create, uh, the SQLite database and a table so that we can store the license plate 670 00:58:26,440 --> 00:58:27,010 data. 671 00:58:27,370 --> 00:58:29,590 So I will resign in protest to like three. 672 00:58:33,880 --> 00:58:36,040 So first we will connect to the. 673 00:58:41,470 --> 00:58:43,630 Light uh database. 674 00:58:47,560 --> 00:58:49,270 Or we will create it. 675 00:59:23,940 --> 00:59:27,690 That will create a cursor object. 676 00:59:30,750 --> 00:59:36,000 Interact with the data base. 677 00:59:45,970 --> 00:59:54,670 So now we will create a table to store the license plate data. 678 59:57.190 --> 1:00:00.970 Let's also execute. 679 1:00:13.670 --> 1:00:17.420 And table if not exists. 680 1:00:28.310 --> 1:00:29.480 So we have. 681 1:01:00.610 --> 1:01:04.120 Then you have the start time and this is in the form of text. 682 1:01:05.290 --> 1:01:07.300 Then you have the end time. 683 1:01:12.340 --> 1:01:17.290 Then we have the license plate like the license plate which are detected. 684 1:01:17.290 --> 1:01:20.060 And this will be in the form of text as well. 685 1:01:25.940 --> 1:01:27.380 So this looks fine. 686 1:01:27.380 --> 1:01:32.810 So now in the meanwhile, like you can see over here, uh, this is ah, all the outputs. 687 1:01:32.810 --> 1:01:36.260 And this is the accumulated JSON file which has been generated. 688 1:01:36.260 --> 1:01:37.400 So this looks good. 689 1:01:39.440 --> 1:01:40.070 Okay. 690 1:01:40.220 --> 1:01:43.020 So now, uh, let's run this. 691 1:01:49.770 --> 1:01:55.800 Now, this will create a SQLite database and a table so that we can where we can store the license plate 692 1:01:55.800 --> 1:01:56.070 data. 693 1:01:56.070 --> 1:01:59.700 So now you can see that, uh, SQLite database have been created. 694 1:01:59.700 --> 1:02:03.750 And so here now we can store our license plate data which you can see over here. 695 1:02:03.750 --> 1:02:08.860 And uh like you can see over here that this file has been created over here. 696 1:02:08.860 --> 1:02:10.510 So, so far, so good. 697 1:02:14.890 --> 1:02:17.890 Now we will save our license plate data to the database. 698 1:02:17.890 --> 1:02:26.530 So now what I will do over here is, uh, you know, write save to. 699 1:02:28.540 --> 1:02:29.590 Database. 700 1:02:32.510 --> 1:02:32.810 Okay. 701 1:02:35.780 --> 1:02:39.110 So the input will be the same which we can find over here. 702 1:02:39.110 --> 1:02:41.270 The exact same input will be passed over here. 703 1:02:43.130 --> 1:02:48.980 So now we'll just write connect Sqlite3 dot connect. 704 1:02:50.900 --> 1:02:53.030 Uh, this is the function created. 705 1:02:55.380 --> 1:02:56.070 Course. 706 1:02:56.070 --> 1:02:56.370 Uh. 707 1:02:58.380 --> 1:03:03.120 One that I saw for plate. 708 1:03:06.120 --> 1:03:06.780 Plates. 709 1:03:11.340 --> 1:03:11.790 Cursor. 710 1:03:11.790 --> 1:03:12.120 Dot. 711 1:03:12.120 --> 1:03:13.080 Execute. 712 1:03:23.710 --> 1:03:24.460 Inside. 713 1:03:38.800 --> 1:03:41.500 Our time and time. 714 1:03:43.130 --> 1:03:43.370 again. 715 1:03:44.180 --> 1:03:44.870 Since. 716 1:03:56.300 --> 1:03:57.620 Then we have values. 717 1:04:25.500 --> 1:04:28.860 I also format like which converted previously as well. 718 1:05:04.910 --> 1:05:09.740 For now, we'll save data to SQL database. 719 1:05:19.740 --> 1:05:26.370 Okay, everything looks fine to me, but, um, let's run this and see how this works. 720 1:05:26.400 --> 1:05:27.180 Start time. 721 1:05:38.550 --> 1:05:41.130 Uh, one thing you need to check. 722 1:06:13.280 --> 1:06:14.960 I think they should match. 723 1:06:18.950 --> 1:06:21.860 Okay, this looks fine. 724 1:06:37.800 --> 1:06:40.320 Yeah, it looks good now. 725 1:06:41.160 --> 1:06:42.540 Just checking it once more. 726 1:06:43.560 --> 1:06:44.640 So let's run this up now. 727 1:07:21.590 --> 1:07:22.580 You can start it. 728 1:07:33.290 --> 1:07:35.540 Okay, so this will take some time. 729 1:07:35.570 --> 1:07:35.870 Okay. 730 1:07:35.870 --> 1:07:36.260 There's a. 731 1:07:41.180 --> 1:07:44.490 Table has no column name and time. 732 1:07:52.770 --> 1:07:53.040 Um. 733 1:07:53.040 --> 1:07:54.570 My mistake. 734 1:08:14.080 --> 1:08:14.830 Understand this. 735 1:08:14.830 --> 1:08:16.540 Hopefully it will work fine now. 736 1:08:37.640 --> 1:08:39.080 There is a mistake as well. 737 1:08:39.080 --> 1:08:40.310 I can see it. 738 1:08:45.020 --> 1:08:45.230 Yeah. 739 1:08:45.230 --> 1:08:49.130 We should have start time and time. 740 1:08:49.160 --> 1:08:49.850 Okay. 741 1:08:53.060 --> 1:08:56.090 I'm now running it again and hopefully this will now work fine. 742 1:09:07.490 --> 1:09:09.260 But let's see how does it goes. 743 1:09:14.000 --> 1:09:14.720 Okay. 744 1:09:15.230 --> 1:09:15.980 Let me remove this. 745 1:09:15.980 --> 1:09:17.120 Previous entries. 746 1:09:25.590 --> 1:09:27.180 On our program has started. 747 1:09:36.570 --> 1:09:37.050 Okay. 748 1:09:37.050 --> 1:09:39.330 And let's see if we have our JSON file. 749 1:09:39.330 --> 1:09:41.880 So now you can see we have our JSON files. 750 1:09:41.880 --> 1:09:42.970 Let's look good. 751 1:09:42.970 --> 1:09:48.670 And if we see now you can see that it's updated as well from the date modified okay. 752 1:09:48.670 --> 1:09:50.980 So now it's hopefully it's working fine. 753 1:09:51.340 --> 1:09:55.990 So let's have some entries and then uh we will uh see our file. 754 1:09:57.040 --> 1:09:58.390 Our database looks like. 755 1:09:59.080 --> 1:10:01.180 So let's continue processing it. 756 1:10:14.150 --> 1:10:16.100 This will take some time still. 757 1:10:16.100 --> 1:10:18.980 I guess it will take much time. 758 1:10:24.980 --> 1:10:29.330 So now you can see as I'm running on CPU machine, the processing is very slow, but if you have a good 759 1:10:29.330 --> 1:10:31.260 GPU the processing will be fast. 760 1:10:31.260 --> 1:10:32.010 Definitely. 761 1:10:43.950 --> 1:10:49.110 So now it's going steady and slow and steady, but it's going smoothly. 762 1:10:50.100 --> 1:10:52.080 So that is what I was expecting. 763 1:10:52.080 --> 1:10:53.430 And it will look good. 764 1:10:53.520 --> 1:10:54.000 Quite good. 765 1:11:01.690 --> 1:11:02.260 Yeah. 766 1:11:03.100 --> 1:11:06.130 So now you can see over here, um, like it's a bit blurred. 767 1:11:06.130 --> 1:11:08.230 So therefore it's painful. 768 1:11:12.730 --> 1:11:16.990 So this is now you can see we have these JSON files till now. 769 1:11:16.990 --> 1:11:22.790 And like you can see over here, our database is being continuously updated like this is our current 770 1:11:22.790 --> 1:11:23.810 database file. 771 1:11:26.360 --> 1:11:28.580 So things look quite smooth and good. 772 1:11:33.500 --> 1:11:34.790 Okay so. 773 1:11:37.010 --> 1:11:40.370 Now to view our SQLite database file. 774 1:11:40.370 --> 1:11:43.440 So I will just write SQLite viewer online. 775 1:11:43.890 --> 1:11:46.620 And over here I will upload this. 776 1:11:46.620 --> 1:11:48.750 So let's wait for it to complete. 777 1:11:48.750 --> 1:11:50.850 And then we will upload our file over here. 778 1:11:55.950 --> 1:11:59.160 By you can see currently we have this I'm just waiting for it to complete. 779 1:11:59.160 --> 1:12:01.470 And then I will show you the file over here. 780 1:12:16.990 --> 1:12:19.330 Currently we have this accumulated data. 781 1:12:19.480 --> 1:12:20.680 That's good as well. 782 1:12:33.350 --> 1:12:34.190 It's a long video. 783 1:12:34.190 --> 1:12:40.670 It's still processing, but we can just pause it over here and currently upload this, our database 784 1:12:40.670 --> 1:12:41.630 file and see. 785 1:12:41.810 --> 1:12:44.840 So I will just go to next to Light Online viewer. 786 1:12:45.110 --> 1:12:49.010 And over here you can see this is my license plate database. 787 1:12:49.010 --> 1:12:50.690 And I will just upload this. 788 1:12:50.690 --> 1:12:54.290 So now you can see these are all the entries we have in our database file. 789 1:12:54.290 --> 1:12:57.510 This is the ID, this is the start time or this is the end time. 790 1:12:57.510 --> 1:13:04.950 And you can see that um like 59 3759, 58 and like this is uh, 22nd time span. 791 1:13:04.950 --> 1:13:11.670 And you can see over here, uh, we have these entries like, like the results were quite good for me 792 1:13:11.670 --> 1:13:11.970 now. 793 1:13:14.100 --> 1:13:18.790 Like, you can see, uh, this is our 22nd again, 22nd time span over here. 794 1:13:18.790 --> 1:13:24.550 And you can see these are the entries which we have found 22nd time span over here as well. 795 1:13:24.550 --> 1:13:28.030 And, uh, that is what looks good to me for now. 796 1:13:28.570 --> 1:13:31.090 Like you can see this is again 22nd time span. 797 1:13:31.090 --> 1:13:33.250 And this is the license plate that has been detected. 798 1:13:33.250 --> 1:13:35.890 And we have just uploaded our database file. 799 1:13:35.890 --> 1:13:39.550 And you can see all the entries over here okay. 800 1:13:39.550 --> 1:13:44.090 3253 like 22nd time span 1333. 801 1:13:44.090 --> 1:13:44.600 Like 26. 802 1:13:44.600 --> 1:13:46.040 Like you can see over here. 803 1:13:46.610 --> 1:13:47.900 So that's good. 804 1:13:47.900 --> 1:13:53.120 So in this tutorial we have learned how we can uh, do uh automatic number plate detection and recognition 805 1:13:53.120 --> 1:13:55.910 using YOLO and uh, OCR. 806 1:13:55.940 --> 1:13:59.390 Plus we have also saved our license plate data into JSON file. 807 1:13:59.390 --> 1:14:04.430 And we have also saved our license plate data into an SQLite database. 808 1:14:04.430 --> 1:14:05.810 So that's all from this tutorial. 809 1:14:05.810 --> 1:14:06.770 Thank you for watching.