Transformer applied in Skeleton-based human action recognition
- download the projet via git:
git clone https://github.com/Annelise2019/DeepLearning_Project.git
- download the dataset NTU-60
And then, this command should be used to build the database for training or evaluation:
python tools/ntu_gendata.py --data_path <path to nturgbd+d_skeletons>
- change the data path in these files
- data_process/skeleton_feeder.py
- config.py
- change the gpu setting in these files
- config.py
- run the model using this command:
python classify.py
- result:
- a folder named "log" will be created to store all the logging traces
- a folder named "checkpoint" will be created to store the model parameters and thus you can retrain your model by changing relative parameters in config.py