[2025INFFUS]STR

STR: Spatio-temporal trajectory representation learning with dual-focus encoder for whole trajectory similarity computation

📄 Link to the paper

Conda Dependencies and Required Packages

• python=3.7.12=hf930737_100_cpython
• pytorch=1.11.0=py3.7_cuda11.3_cudnn8.2.0_0
• torchaudio=0.11.0=py37_cu113
• torchvision=0.12.0=py37_cu113
• pandas==1.2.4
• numpy==1.21.5
• trajectory-distance==1.0
• tensorboard==2.5.0

Running

python main.py --config=config.yaml --gpu=0

The relevant parameters can be modified in the config.yaml file.

Citation

If you find anything in this repository useful to your research, please cite our paper :) We sincerely appreciate it.

@article{li2025str,
  title={STR: Spatio-temporal trajectory representation learning with dual-focus encoder for whole trajectory similarity computation},
  author={Li, Mengqiu and Niu, Xinzheng and Zhu, Jiahui and Fournier-Viger, Philippe and Wu, Youxi},
  journal={Information Fusion},
  pages={103231},
  year={2025},
  publisher={Elsevier}
}