Cones_tracking_model

Model was made as a task in the science club : AGH_RACING

Configuration:

⚫STEP 1

• Clone YOLOv7

  $ git clone https://github.com/WongKinYiu/yolov7
  $ cd yolov7
  $ pip install -qr requirements.txt
  

⚪STEP 2

• Move file: 'cones.yaml' to yolov7/data

• Prepare your dataset. Data is in the Supervisely format originally, but yolov7 requires YOLOv7 PyTorch TXT format.

  You can automatically preprocess data using execute_all.py script in the repository:

  python execute_all.py -d {path_to_your_dataset} -y {path_to_your_yolo_folder} -i {path_to_your_images_folder} -a {path_to_your_annotations_folder}
  

  or by using the scripts separately:

  python3 _1_parse_annotations.py -i {path_to_your_images_folder} -d {path_to_your_dataset} -a {path_to_your_annotations_folder} -y {path_to_your_yolo_folder}
  python3 _2_copy_images_to_single_directory.py -i {path_to_your_images_folder} -d {path_to_your_dataset} -a {path_to_your_annotations_folder} -y {path_to_your_yolo_folder}
  python3 _3_partition_data.py -i {path_to_your_images_folder} -d {path_to_your_dataset} -a {path_to_your_annotations_folder} -y {path_to_your_yolo_folder}
  

The final sample result in your YOLOv7 folder should look like this:

          yolov7
          |
          ├───images
          |   ├── test
          |   |   └─────────  amz_00000.jpg
          |   |               amz_00000.txt
          |   |               amz_00003.jpg
          |   |               amz_00003.txt
          |   |               amz_00006.jpg
          |   |               amz_00006.txt
          |   |               amz_00011.jpg
          |   |               amz_00011.txt
          |   |               amz_00013.jpg
          |   |               amz_00013.txt
          |   ├── train
          |   |   └─────────  amz_00001.jpg
          |   |               amz_00001.txt
          |   |               amz_00002.jpg
          |   |               amz_00002.txt
          |   |               amz_00004.jpg
          |   |               amz_00004.txt
          |   |               amz_00005.jpg
          |   |               amz_00005.txt
          |   |               amz_00007.jpg
          |   |               amz_00007.txt
          |   |     
          |   |     
          |   └── val
          |       └─────────  amz_00022.jpg
          |                   amz_00022.txt
          |                   amz_00033.jpg
          |                   amz_00033.txt
          |                   amz_00051.jpg
          |                   amz_00051.txt
          |                   amz_00063.jpg
          |                   amz_00063.txt
          |                   amz_00069.jpg
          |                   amz_00069.txt
          |   
          └── data
              └─── cones.yaml 

⚫STEP 3

Run 'train_command.txt' on your automatically moved and preprocessed dataset or use one of the following commands:

  ⚪ CPU training
  $ python train.py --img 640 --cfg cfg/training/yolov7-tiny.yaml  --batch-size 1 --epochs 10 --data cones.yaml --weights '' --workers 1 --device cpu --name cones


  ⚫ Single GPU training

    train p5 models
  $ python train.py --workers 8 --device 0 --batch-size 32 --data cones.yaml --img 640 640 --cfg cfg/training/yolov7.yaml --weights '' --name yolov7 --hyp data/hyp.scratch.p5.yaml

    train p6 models
  $ python train_aux.py --workers 8 --device 0 --batch-size 16 --data cones.yaml --img 1280 1280 --cfg cfg/training/yolov7-w6.yaml --weights '' --name yolov7-w6 --hyp data/hyp.scratch.p6.yaml


  ⚪ Multiple GPU training

    train p5 models
  $ python -m torch.distributed.launch --nproc_per_node 4 --master_port 9527 train.py --workers 8 --device 0,1,2,3 --sync-bn --batch-size 128 --data cones.yaml --img 640 640 --cfg cfg/training/yolov7.yaml --weights '' --name yolov7 --hyp data/hyp.scratch.p5.yaml

    train p6 models
  $ python -m torch.distributed.launch --nproc_per_node 8 --master_port 9527 train_aux.py --workers 8 --device 0,1,2,3,4,5,6,7 --sync-bn --batch-size 128 --data cones.yaml --img 1280 1280 --cfg cfg/training/yolov7-w6.yaml --weights '' --name yolov7-w6 --hyp data/hyp.scratch.p6.yaml

Model's basic documentation:

• Dataset consists of 24 GB photos of cones in Supervisely format. Training set consists 80% of all photos and testing set consists 20% of all photos.

Model properties:

Dataset is available under the following link:

• https://www.fsoco-dataset.com/download