Using computer vision to counting japanese coins
Technique: object detection Consider each coin is an object that need to be detect.
Model to train: RetinaNet
Japanese coin image dataset: https://drive.google.com/drive/folders/1tR5S4su0RXdCuyNFc6CDHXaI4zxxdYl7?usp=sharing
Model weight file: https://drive.google.com/drive/folders/1cL-QQc4Oc2R-cz-cY10eYWLDAlmvXxs1?usp=sharing
Test some images after training
To count coins in an image run this command:
python count.py --model_dir=path_to_trained_model --image_dir=path_to_image_file