LicensePlateDetection/setup.txt at main · thanhan11hz/LicensePlateDetection · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
# Set up môi trường Python trên Linux
   apt update

   # Với YOLO
   apt install -y python3 python3-pip python3-venv

   # Với Faster RCNN, RetinaNet, RTMDet
   apt update
   apt install software-properties-common -y
   add-apt-repository ppa:deadsnakes/ppa
   sudo apt install python3.9 python3.9-venv python3.9-dev -y

   python3.9 -m venv venv
   source venv/bin/activate

# Clone thư mục từ drive
   apt install -y rclone
   rclone config
   n
   drive
   drive
   Enter
   Enter
   Enter
   1
   Enter
   Enter
   n
   y
   Dang nhap
   n
   y
   q

# Tải thư mục từ drive

   mkdir -p /root/yolo_dataset
   rclone copy "drive:/Đại học/Năm 2/images.zip" /root/yolo_dataset -P
   rclone copy "drive:/Đại học/Năm 2/YOLO_labels.zip" /root/yolo_dataset -P
   rclone copy "drive:/resnet50-0676ba61.pth" /root/COCO_dataset -P

# Unzip thư mục
   apt install -y unzip
   unzip __tenfile

# Cài các package để train model
   source /root/venv/bin/activate
   - Pytorch => Xem GPU để biết tải bản torch cho CUDA nào
      pip uninstall torch torchvision torchaudio -y
      pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124
      pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
   - Ultralytics (YOLO)
      pip install -U ultralytics
      yolo checks
      apt install -y libgl1
      apt install -y libglib2.0-0

      + Chạy test
      yolo detect val \
      model=work_dirs/license_plate/yolov8/weights/best.pt \
      data=datasets/license_plate.yaml \
      imgsz=640 \
      save_json=True
   - MMDetection (Faster-RCNN, RetinaNet, RTMDet)
      conda install pytorch==1.9.1 torchvision==0.10.1 torchaudio==0.9.1 cudatoolkit=11.1 -c pytorch -c conda-forge
      pip install mmengine==0.10.7
      pip install mmcv==2.0.0 -f https://download.openmmlab.com/mmcv/dist/cu111/torch1.9.0/index.html
      pip install "mmdet>=3.0.0"

      + Clone MMDetection về
         git --version
         * Nếu không có git trên server
            sudo apt update
            sudo apt install git -y
      + Copy file configs
         cp configs/faster_rcnn/faster-rcnn_r50_fpn_1x_coco.py \ configs/license_plate
         cp configs/retinanet/retinanet_r50_fpn_1x_coco.py \ configs/license_plate
         cp configs/rtmdet/rtmdet_s_8xb32-300e_coco.py \ configs/license_plate

      + Train từ đầu
         python tools/train.py \ configs/license_plate/faster_rcnn_r50_1x_coco.py \ --work-dir work_dirs/license_plate/faster_rcnn
         python tools/train.py \ configs/license_plate/retinanet_r50_fpn_1x_coco.py
         python tools/train.py \ configs/license_plate/rtmdet_s_8xb32-300e_coco.py

      + Train từ checkpoint nếu bị crash server
         python tools/train.py \ configs/license_plate/faster_rcnn_r50_1x_coco.py --resume
         python tools/train.py \ configs/license_plate/retinanet_r50_fpn_1x_coco.py --resume
         python tools/train.py \ configs/license_plate/rtmdet_s_8xb32-300e_coco.py --resume

      + Chạy test
      python tools/test.py \
      configs/license_plate/faster_rcnn_r50_fpn_bs32_20e.py \
      work_dirs/faster_rcnn_r50_fpn_bs32_20e/best_bbox_mAP.pth \
      --out work_dirs/faster_rcnn_r50_fpn_bs32_20e/preds.json


# Note: Cấu trúc thư mục khi train bằng MMDetection
mmdet_project/
│
├── mmdetection/
├── configs/
│   ├── _base_/
│   ├── faster_rcnn/
│   ├── retinanet/
│   ├── rtmdet/
│   └── license_plate/
│       ├── faster_rcnn_r50_fpn_bs32_20e.py <- File config của mình
│       ├── retinanet_r50_fpn_bs32_20e.py <- File config của mình
│       └── rtmdet_s_bs32_20e.py <- File config của mình
│
├── data/
│   └── license_plate/
│       ├── train/
│       │   ├── 0001.jpg
│       │   └── ...
│       ├── val/
│       │   ├── 1001.jpg
│       │   └── ...
│       └── annotations/
│           ├── train.json
│           └── val.json
│
└── work_dirs/
    └── faster_rcnn_bs32_20e/