Home

Awesome

DSTA-Net

Decoupled Spatial-Temporal Attention Network for Skeleton-Based Action-Gesture Recognition in ACCV2020

Result

A little different with paper due the reimplementation.

Data Preparation

Training & Testing

Change the config file depending on what you want.

`python train_val_test/train.py --config ./config/shrec/shrec_dstanet_14.yaml`

Train with decoupled modalities by changing the 'num_skip_frame'(None to 1 or 2) option and 'decouple_spatial'(False to True) option in config file and train again.

Then combine the generated scores with:

`python train_val_test/ensemble.py`
 

Citation

Please cite the following paper if you use this repository in your reseach.

@inproceedings{dstanet_accv2020,  
      title     = {Decoupled Spatial-Temporal Attention Network for Skeleton-Based Action-Gesture Recognition},  
      author    = {Lei Shi and Yifan Zhang and Jian Cheng and Hanqing Lu},  
      booktitle = {ACCV},  
      year      = {2020},  
}

Contact

For any questions, feel free to contact: lei.shi@nlpr.ia.ac.cn