STGlow: A Flow-based Generative Framework with Dual Graphormer for Pedestrian Trajectory Prediction

Rongqin Liang, Yuanman Li, Jiantao Zhou, and Xia Li

Our STGlow network architecture:

Installation

Dependencies

• Python 3.8
• pytorch 1.11.0
• cuda 11.3
• Ubuntu 20.04
• RTX 3090
• Please refer to the "requirements.txt" file for more details.

Training

Users can train the STGlow models on ETH/UCY or SDD dataset easily by runing the following command:

For ETH/UCY:

python tools/train_for_eth_ucy.py 

For SDD:

python tools/train_for_sdd.py 

Inference

Users can test the STGlow models on ETH/UCY or SDD dataset easily by runing the following command:

For ETH/UCY:

python tools/test_for_eth_ucy.py 

For SDD:

python tools/test_for_sdd.py 

Note that our project is developed based on the code of BiTraP: Bi-directional Pedestrian Trajectory Prediction with Multi-modal Goal Estimation.

Citation

If you found the repo is useful, please feel free to cite our papers:

@article{liang2022stglow,
      title={STGlow: A Flow-based Generative Framework with Dual Graphormer for Pedestrian Trajectory Prediction}, 
      author={Rongqin Liang and Yuanman Li and Jiantao Zhou and Xia Li},
      journal={arXiv preprint arXiv:2211.11220}
      year={2022}
}