We propose a variant of GCNs to leverage the powerful self-attention mechanism to adaptively sparsify a complete action graph in the temporal space. Our method could dynamically attend to important past frames and construct a sparse graph to apply in the GCN framework, well-capturing the structure information in action sequences. The paper can be found on arXiv here!. ECCV 2020 accepted.
If you use this code or these models, please cite the following paper:
@article{yu2020structure,
title={Structure-Aware Human-Action Generation},
author={Yu, Ping and Zhao, Yang and Li, Chunyuan and Chen, Changyou},
journal={arXiv preprint arXiv:2007.01971},
year={2020}
}
Fig. 1: The overall framework of the proposed method.
Fig.1 illustrates the overall framework of our model for action generation. It follows the GAN framework of video generation, which consists of an action generator G and a dual discriminator: one video-based discriminator D and one frame-based discriminator D_F.
python3 train.py
Please drop me a line or submit an issue to this Github page if you have any questions.