pip install -r requirements.txt
Dataset should be folder data
# We use all the train data and 2/3 of the val data as the final training set, and the remaining val data as the validation dataset.
python data_arrange.py
# We have usd a pre-trained model based on ImageNet provided by timm.
python main.py
# When the training is over, you can copy the checkpoint.pth.tar in reulst to the root directory and use it for testing.
python test.py