Introduction

This is the source code of our CFSFNet paper "Channel-level Feature Selection and Fusion Network for Visible-infrared Person Re-identification". Please cite the following paper if you use our code.

Dependencies

• Python 3.7.11
• PyTorch 1.10.1
• cudatoolkit 11.3.58
• torchvision 0.11.2

Data Preparation

Download the SYSU-MM01 dataset and the RegDB dataset, and place them to /home/seayuan/Dataset/ folders.

Usage

• Start training by executing the following commands.

1.For SYSU-MM01 dataset:

Train: python train.py --dataset sysu --gpu 0

Test: python test.py --dataset sysu --model_path 'path' --gpu 0

2.For RegDB dataset:

python train.py --dataset regdb --gpu 0

For any questions, feel free to contact us.