Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:01,050 --> 00:00:04,980
Okay, then we will do training of these seven on custom objects next.
2
00:00:04,980 --> 00:00:08,220
So the dataset is set up and the configuration file is created.
3
00:00:08,580 --> 00:00:12,540
If are doing training, we have to download Pre-trained weights for transfer learning.
4
00:00:12,570 --> 00:00:14,450
Visit the following repositories.
5
00:00:14,460 --> 00:00:19,560
Scroll down after the download YOLO of seven Training from the Transfer Learning section, click on
6
00:00:19,560 --> 00:00:20,580
the following link.
7
00:00:24,040 --> 00:00:25,870
Wait until the download is finished.
8
00:00:37,670 --> 00:00:41,150
When you finish, navigate to the weight file in the downloads folder.
9
00:00:41,720 --> 00:00:45,410
After that, move the words to the all of 70 plus route folder.
10
00:00:46,760 --> 00:00:49,160
Click the file, then press control X.
11
00:00:53,750 --> 00:00:57,650
Place in the all of seven GP use route followed by pressing control fee.
12
00:01:02,700 --> 00:01:06,480
But before doing the training, we will explain some arguments that can be used.
13
00:01:06,510 --> 00:01:08,370
First, there is the worker's argument.
14
00:01:08,400 --> 00:01:12,300
This argument is the number of processes that generate parties in parallel.
15
00:01:12,330 --> 00:01:14,580
This is an example of its application.
16
00:01:15,720 --> 00:01:17,460
Next is the bed size argument.
17
00:01:17,490 --> 00:01:21,040
This argument is the number of images processed before updating the model.
18
00:01:21,060 --> 00:01:23,190
This is an example of its application.
19
00:01:24,680 --> 00:01:30,470
If the CPU used for training has relatively small set workers to zero and reduce the bad size.
20
00:01:33,610 --> 00:01:35,110
Makes this the device argument.
21
00:01:35,140 --> 00:01:37,000
This argument is kill the device.
22
00:01:37,660 --> 00:01:39,850
This is an example of its application.
23
00:01:41,670 --> 00:01:44,520
If using Cipro, you write Cipro, you and device.
24
00:01:44,850 --> 00:01:46,350
Next is the data argument.
25
00:01:46,380 --> 00:01:51,570
This argument is a data file that contains the number of classes that Cipro and class name.
26
00:01:52,730 --> 00:01:54,890
This is an example of its application.
27
00:01:55,580 --> 00:01:57,440
Next up is the IMT argument.
28
00:01:57,470 --> 00:01:59,990
This argument is the size of the image to be trained.
29
00:02:01,640 --> 00:02:03,770
This is an example of its application.
30
00:02:05,750 --> 00:02:07,370
Next is the CFD argument.
31
00:02:07,400 --> 00:02:09,710
This argument is a configuration file.
32
00:02:11,670 --> 00:02:13,770
This is an example of its application.
33
00:02:14,660 --> 00:02:16,160
Next week's argument.
34
00:02:16,160 --> 00:02:20,240
This argument is a wedge file that is uses Pre-trained words for transform learning.
35
00:02:20,660 --> 00:02:22,880
This is an example of its application.
36
00:02:24,720 --> 00:02:25,850
Next name argument.
37
00:02:25,860 --> 00:02:28,320
This argument is the name of the model to be trained.
38
00:02:31,780 --> 00:02:33,850
This is an example of its application.
39
00:02:34,600 --> 00:02:35,950
Next hip argument.
40
00:02:35,980 --> 00:02:38,770
This argument is a YOLO of seven separate parameter.
41
00:02:41,810 --> 00:02:43,940
This is an example of its application.
42
00:02:46,530 --> 00:02:48,120
Next is the epochs argument.
43
00:02:48,150 --> 00:02:53,010
This argument is the number of times the learning algorithm will work to process the entire dataset.
44
00:02:54,700 --> 00:02:56,800
This is an example of its application.
45
00:02:59,640 --> 00:03:03,090
The first step in performing the training is to launch the Anaconda prompt.
46
00:03:03,180 --> 00:03:06,000
Press the Windows button, then enter Anaconda.
47
00:03:06,030 --> 00:03:07,500
Click the Anaconda prompt.
48
00:03:13,260 --> 00:03:17,310
Then activate the all of seven CPU environment using the command.
49
00:03:17,340 --> 00:03:20,460
Activate yolo 574 for and V.
50
00:03:22,370 --> 00:03:23,210
Press internal.
51
00:03:25,280 --> 00:03:28,340
Then navigate to the YOLO seven zip use route folder.
52
00:03:33,330 --> 00:03:39,810
We will do training with the face mask dataset, the Nvidia GeForce GTX 1650 TI with four gigabytes
53
00:03:39,810 --> 00:03:40,440
of GPU.
54
00:03:40,440 --> 00:03:45,660
RAM is used for training used to command below fight and train dog p y.
55
00:03:47,690 --> 00:03:49,220
In workers set zero.
56
00:03:50,620 --> 00:03:52,420
In the bedside we write for.
57
00:03:52,450 --> 00:03:56,290
You can increase the bed size value if you're using a GPU with more RAM.
58
00:03:58,360 --> 00:03:59,710
On device zero.
59
00:04:05,970 --> 00:04:06,890
In the data, right?
60
00:04:06,900 --> 00:04:08,970
The data file that was previously created.
61
00:04:17,480 --> 00:04:19,760
At IMDB, we read 640.
62
00:04:23,240 --> 00:04:27,020
In the CFG write the configuration file that was previously created.
63
00:04:34,400 --> 00:04:34,990
In width.
64
00:04:35,000 --> 00:04:37,640
We use yolo v seven training as initial weights.
65
00:04:47,330 --> 00:04:48,650
In the name we write YOLO.
66
00:04:48,650 --> 00:04:50,030
Five seven Face mask.
67
00:05:01,320 --> 00:05:04,590
In hit use helps create custom file in the data folder.
68
00:05:10,330 --> 00:05:11,020
In epochs.
69
00:05:11,020 --> 00:05:12,100
We write 300.
70
00:05:17,330 --> 00:05:18,890
Press enter to start training.
71
00:05:27,240 --> 00:05:29,370
The following is the training process.
72
00:05:53,110 --> 00:05:53,440
It's a.
73
00:05:53,800 --> 00:05:56,500
We'll calculate the MLP to get the best weights.
74
00:06:22,490 --> 00:06:26,840
If you want to stop training before the specified epochs, you can press control C.
75
00:06:28,560 --> 00:06:34,140
The future training results once Windows Explorer and navigate to the YOLO v seven Use route folder.
76
00:06:35,660 --> 00:06:38,540
The training results are stored in the runs train.
77
00:06:39,570 --> 00:06:40,440
Model name.
78
00:06:45,610 --> 00:06:48,580
The training without weights file is stored in the weights folder.
79
00:06:53,660 --> 00:06:59,060
That's not pretty is the words with the highest MLP value last the p t is the weights of the last epoch.
80
00:07:17,460 --> 00:07:20,160
Next we can see the training graph using the tensor board.
81
00:07:23,260 --> 00:07:24,550
Use the command below.
82
00:07:24,580 --> 00:07:25,420
Sensible.
83
00:07:27,620 --> 00:07:27,900
That's.
84
00:07:27,940 --> 00:07:28,880
That's floor zero.
85
00:07:40,540 --> 00:07:41,590
One strain.
86
00:07:44,510 --> 00:07:45,290
First internal.
87
00:07:48,390 --> 00:07:51,930
Copy the following link by blocking like this, then right click.
88
00:07:55,500 --> 00:07:56,580
Open your browser.
89
00:07:57,750 --> 00:07:58,920
Taste the cup and link.
90
00:08:09,880 --> 00:08:12,310
The following is a graph of the training results.
91
00:08:15,480 --> 00:08:16,230
For performance.
92
00:08:16,230 --> 00:08:20,730
You can pay attention to the map graph of mean average precision, the high of the better.
93
00:08:36,280 --> 00:08:39,250
In the next video, we will explain how to continue training.
94
00:08:39,490 --> 00:08:40,270
See you then.
7777
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.