Cards image classification

1. How to prepare dataset

• Download dataset from here
• Then

mkdir cards-image-datasetclassification
mv cards-image-datasetclassification.zip cards-image-datasetclassification
cd cards-image-datasetclassification
unzip -q cards-image-datasetclassification.zip

2. Set up environment

conda create -n myenv python=3.9
conda activate myenv
pip install -r requirements.txt

3.Training process

Training script:

python main.py --train-folder ${train_folder} --test-folder ${valid_folder} --batch-size ${bs}  --learning-rate ${lr} --epochs ${epochs}

Example:

python main.py -bs 16 --epochs 1

4. Serve model

After training, we have the model saved in model folder. We can run FastAPI to serve the model.

cd app
python main.py

Then open browser and go to localhost:8088/docs to see the API documentation.

5. Set up Grafana dashboard

docker compose -f grafana/docker-compose.yml up -d

Then open browser and go to localhost:3000 to see the Grafana dashboard.

6. Additonal

Set up pre-commit

pre-commit install
pre-commit run --all-files