STIGPN

Space-Time Interaction Graph Parsing Networks for Human-Object Interaction Recognition，ACM MM'21

Installation

1. Clone this repository.

   git clone https://github.com/GuangmingZhu/STIGPN.git
   

2. Install Python dependencies:

   pip install -r requirements.txt
   

Prepare Data

1. Follow here to prepare the original data of CAD120 dataset in CAD120/datasets folder.
2. You can also download the data we have processed directly from here.
3. We also provide some checkpoints to the trained models. Download them here and put them in the checkpoints folder

Training

For the CAD120 dataset: python train_CAD120.py --model VisualModelV python train_CAD120.py --model SemanticModelV

Testing

For the CAD120 dataset: python eval_CAD120.py

Citation

If you use our annotations in your research or wish to refer to the baseline results, please use the following BibTeX entry.

@inproceedings{wang2021spatio,
  title={Spatio-Temporal Interaction Graph Parsing Networks for Human-Object Interaction Recognition},
  author={Wang, Ning and Zhu, Guangming and Zhang, Liang and Shen, Peiyi and Li, Hongsheng and Hua, Cong},
  booktitle={Proceedings of the 29th ACM International Conference on Multimedia},
  pages={4985--4993},
  year={2021}
}