Mutual Distillation Driven Dual-Space Matching for Visible–Infrared Person Re-Identification

Welcome to use the code from our paper "Mutual Distillation Driven Dual-Space Matching for Visible–Infrared Person Re-Identification".

Environment

• Python 3.10+
• PyTorch >= 2.0.1

Dataset

SYSU-MM01, RegDB, and LLCM

Preprocessing

Preprocess SYSU-MM01 and LLCM datasets for faster training:

python pre_process_sysu.py
python pre_process_llcm.py

Training and Testing

We provide convenient shell scripts for training and testing. You can directly follow the scripts or run the commands manually.

# Using the script
bash train.sh

Requirements

we use single RTX4090 24G GPU for training and evaluation.

Acknowledgements

The code is based on DNS, CAJ, and DEEN. We thank the authors for open-sourcing their code.

Contact

If you have any questions, don't hesitate to contact me via dongcanliu@stu.cqut.edu.cn.