Skip to content

Latest commit

 

History

History
45 lines (34 loc) · 1.14 KB

README.md

File metadata and controls

45 lines (34 loc) · 1.14 KB
Raw

Exploring Dynamic Context for Multi-path Trajectory Prediction

DCENet Structure

DCENet

Requiements

  • python3
  • keras-gpu 2.3.1
  • tensorflow 2.1.0
  • numpy ...
pip install -r requiements.txt

Data Preparation

  1. download raw data from directory /WORLD H-H TRAJ, and save in /processed_data
  2. run /scripts/trainer.py by setting arg.preprocess==True for data processing. Note: set arg.preprocess==False when you have already the processed data when you run /scripts/trainer.py to save time.

Test

You can get the results as reported in the paper using our pretrained model.

  1. Download pretrained model from /models/best.hdf5

Train

You also can train from sratch by /scripts/trainer.py

Bibtex

If you find our work useful for you, please cite it as:

@article{cheng2020exploring,
  title={Exploring Dynamic Context for Multi-path Trajectory Prediction},
  author={Cheng, Hao and Liao, Wentong and Tang, Xuejiao and Yang, Michael Ying and Sester, Monika and Rosenhahn, Bodo},
  journal={arXiv preprint},
  year={2020}
}