Home

Awesome

Towards More Practical Group Activity Detection:<br> A New Benchmark and Model

Dongkeun Kim, Youngkil Song, Minsu Cho, Suha Kwak

Project Page | Paper

Overview

This work introduces the new benchmark dataset, Café, and a new model for group activity detection (GAD).

Requirements

Conda environment installation

conda env create --file environment.yml

conda activate gad

pip install torch==1.7.1+cu110 torchvision==0.8.2+cu110 -f https://download.pytorch.org/whl/torch_stable.html

Install additional package

sh scripts/setup.sh

Download datasets

Download Café dataset from: <br/>

https://cvlab.postech.ac.kr/research/CAFE/

Download trained weights

sh scripts/download_checkpoints.sh

or from: <br/>
https://drive.google.com/file/d/1W_2gkzARCzSdK8Db4G4pkzN3GrJTYo8R/view?usp=drive_link

Run test scripts

Run train scripts

File structure

├── Dataset/
│     └── cafe/
│           └── gt_tracks.pkl
├── dataloader/
├── evaluation/
│     └── gt_tracks.txt
├── label_map/
├── models/
├── scripts/
└── util/
train.py
test.py
environment.yml
README.md

Citation

If you find our work useful, please consider citing our paper:

@article{kim2023towards,
  title={Towards More Practical Group Activity Detection: A New Benchmark and Model},
  author={Kim, Dongkeun and Song, Youngkil and Cho, Minsu and Kwak, Suha},
  journal={arXiv preprint arXiv:2312.02878},
  year={2023}
}

Acknowledgement

This work was supported by the NRF grant and the IITP grant funded by Ministry of Science and ICT, Korea (RS-2019-II191906, IITP-2020-0-00842, NRF-2021R1A2C3012728, RS-2022-II220264).