On Retrospecting Human Dynamics with Attention

This is the source code for the paper

M. Dong and C. Xu, “On retrospecting human dynamics with attention,”in Proceedings of the Twenty-Eighth International Joint Conferenceon Artificial Intelligence, IJCAI-19.International Joint Conferenceson Artificial Intelligence Organization, 7 2019, pp. 708–714.

The paper proposes a retropsection module with attention for human motion prediction. For more details, please refer to our paper on ResearchGate.

Bibtex:

@inproceedings{ijcai2019-100,
  title     = {On Retrospecting Human Dynamics with Attention},
  author    = {Dong, Minjing and Xu, Chang},
  booktitle = {Proceedings of the Twenty-Eighth International Joint Conference on
               Artificial Intelligence, {IJCAI-19}},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},             
  pages     = {708--714},
  year      = {2019},
  month     = {7},
  doi       = {10.24963/ijcai.2019/100},
  url       = {https://doi.org/10.24963/ijcai.2019/100},
}

Dependencies

• h5py -- to save samples
• Tensorflow 1.2 or later.

Train

Run:
To train a model from scratch, run:

python src/translate.py --omit_one_hot --sub_loss_weight 0.5 --reconstruction_len 4 --anchor_len 4 --experiments 0

Visualize:
To visualize the predicted results, run:

python src/forward_kinematics.py

Acknowledgments

The pre-processed human 3.6m dataset and some of our evaluation code was ported/adapted from SRNN by @asheshjain399 and RRNN by @una-dinosauria.

Licence

MIT