HIT

This project is the official implementation of our paper Holistic Interaction Transformer Network for Action Detection (WACV 2023), authored by Gueter Josmy Faure, Min-Hung Chen and Shang-Hong Lai.

Updates

• (03/06/2023) We have added the code to train/test on AVA here. Any issues about AVA, please open them from the other repo.

Demo Video

Installation

You need first to install this project, please check INSTALL.md

Data Preparation

To do training or inference on J-HMDB, please check DATA.md for data preparation instructions. Instructions for other datasets coming soon.

Model Zoo

Please see MODEL_ZOO.md for downloading models.

Training and Inference

To do training or inference with HIT, please refer to GETTING_STARTED.md.

Citation

If this project helps you in your research or project, please cite this paper:

@InProceedings{Faure_2023_WACV,
    author    = {Faure, Gueter Josmy and Chen, Min-Hung and Lai, Shang-Hong},
    title     = {Holistic Interaction Transformer Network for Action Detection},
    booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
    month     = {January},
    year      = {2023},
    pages     = {3340-3350}
}

Acknowledgement

We are very grateful to the authors of AlphAction for open-sourcing their code from which this repository is heavily sourced. If your find this research useful, please consider citing their paper as well.

@inproceedings{tang2020asynchronous,
  title={Asynchronous Interaction Aggregation for Action Detection},
  author={Tang, Jiajun and Xia, Jin and Mu, Xinzhi and Pang, Bo and Lu, Cewu},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  year={2020}
}