Set up experiments on deep learning models with UBK data
pip install torch
pip install torchsampler
Data should be processed first and put aside in dataset
directory. The raw data and processed data are made available on request only.
We provide MLP, AlexNet, Vgg16, Resnet18, Transformer, and Swin-transformer models for evaluation, you can conduct experiments and select
models by --config-file
argument shown as below. The corresponding config files are listed in config
directory.
# run with the demo data in git repo
python run.py -c config/alexnet_demo.yaml
# cd path/to/UBKProj
python run.py --config-file config/vgg16_1d.yaml
python run.py --config-file config/vgg16_1d.yaml TRAIN.RESUME /path/to/weight
python run.py --config-file config/vgg16_1d.yaml --test-only TEST.WEIGHT /path/to/weight