GroupRec

[ICCV2023] Reconstructing Groups of People with Hypergraph Relational Reasoning (GroupRec)

The official code for ICCV 2023 paper "Reconstructing Groups of People with Hypergraph Relational Reasoning"
Buzhen Huang, Jingyi Ju, Zhihao Li, Yangang Wang
[Project] [Paper] [Video]

Installation

Create conda environment and install dependencies.

conda create -n grouprec python=3.8
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu111 # install pytorch
pip install -r requirements.txt

Getting Started

Step1:
Download the official SMPL model from SMPLify website and put it in data/SMPL_NEUTRAL.pkl.

Step2:
Download trained models from Baidu Netdisk and put them in data.

Step3:
Run demo for 3D pose estimation.

python demo.py --config cfg_files/demo.yaml

Run demo for SMPL estimation.

python demo.py --config cfg_files/demo_smpl.yaml

Train

You can download the training data from Baidu Netdisk and place it in the data directory. Due to licensing restrictions, we cannot provide the Human3.6M dataset. However, you may process it into the same format as our provided data.

python main.py --config cfg_files/config.yaml

Pseudo Dataset

We provide pseudo annotations for Panda dataset (Detection and MOT). You may also need to download image files from their official websites.

[Annotations] [Detection Image] [MOT Image]

TODOS:

• Demo code for pose estimation
• Demo code for SMPL estimation
• Training code release

Citation

If you find this code useful for your research, please consider citing the paper.

@inproceedings{grouprec,
title={Reconstructing Groups of People with Hypergraph Relational Reasoning},
author={Huang, Buzhen and Ju, Jingyi and Li, Zhihao and Wang, Yangang},
booktitle={ICCV},
year={2023},
}

Acknowledgments

Some of the code are based on the following works. We gratefully appreciate the impact it has on our work.
CLIFF
ByteTrack
LoCO
YOLOX