MST-GCN

This is the official implemntation for "Multi-scale spatial temporal graph convolutional network for skeleton-based action recognition" AAAI-2021 (pdf)

Requirements

Data Preparation

NTU RGB+D dataset

• Download the raw data of NTU RGB+D.
• Preprocess the data with python data_gen/ntu_gendata.py
• Generate the bone data with python data_gen/gen_bone_data.py
• Generate the motion data with python data_gen/gen_motion_data.py

Kinetics-Skeleton

• Download the data from GoogleDrive provided by st-gcn
• Preprocess the data with python data_gen/kinetics_gendata.py
• Generate the bone data with python data_gen/gen_bone_data.py
• Generate the motion data with python data_gen/gen_motion_data.py

Training-Testing

Change the config file depending on what you want.

# train on NTU RGB+D xview joint stream
$ sh run.sh 0,1,2,3 4 2022 0
# or
$ CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --nproc_per_node=4 --master_port=2022 main.py --config config/ntu/train_joint_amstgcn_ntu.yaml

Citation

Please cite our paper if you find this repository useful in your resesarch:

@inproceedings{chen2021multi,
  title={Multi-scale spatial temporal graph convolutional network for skeleton-based action recognition},
  author={Chen, Zhan and Li, Sicheng and Yang, Bing and Li, Qinghan and Liu, Hong},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={35},
  number={2},
  pages={1113--1122},
  year={2021}
}

Acknowledgement

The framework of our code is extended from the following repositories. We sincerely thank the authors for releasing the codes.

• The framework of our code is based on st-gcn and 2s-agcn.

Licence

This project is licensed under the terms of the MIT license.

Contact

For any questions, feel free to contact: [email protected] or [email protected]