pstephenwille / yolov8-playingcards

training ML to recognize playing cards

Playing Card Detection with YOLOv8

Guide:

• Having Fun with YOLOv8: How Good Your Model in Detecting Playing Card?.

Steps

1. install nvidia-container-toolkit https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html
2. build the dockerfile: docker build -t cuda-12-1:latest .

3.  docker run \
    -p 8888:8888 \
    -v $PWD:/home \
    --rm --runtime=nvidia \
    --gpus all \
    --ipc=host \
    cuda-12-1:latest
   

4. verify container is running on GPU:

   python3
   import torch
   dev = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
   

5. check localhost:8888
6. run the notebook

todo

1. increase dataset to obtain >90% confidence in all cards

Notes

ultralytics missing deps: 'tf_keras', 'sng4onnx>=1.0.1', 'onnx_graphsurgeon>=0.3.26', 'onnx>=1.12.0', 'onnx2tf>1.17.5,<=1.22.3', 'onnxslim>=0.1.31', 'tflite_support', 'onnxruntime'