Dual-level Aggregation Network For Video-based Visible-Infrared Group Re-Identification

This is the official repo of our PR paper 'Dual-level Aggregation Network For Video-based Visible-Infrared Group Re-Identification'.

Code: Train DAN by

python train.py --gpu your_gpu_id --root your_dataset_directory

Dataset: To access the video-based visible-infrared group re-id dataset VVIG, please send a signed agreement to silverpride@foxmail.com, and we will send you the download link.

Citation: Please cite the following paper if you use this repository in your research work.

@article{XIONG2026112079,
title = {Dual-level aggregation network for video-based visible-infrared group re-identification},
journal = {Pattern Recognition},
volume = {170},
pages = {112079},
year = {2026},
author = {Jianghao Xiong and Xiaohua Xie and Jian-Huang Lai},
}

Acknowledgment: The code is based on MITML.

@inproceedings{lin2022learning,
 title={Learning Modal-Invariant and Temporal-Memory for Video-Based Visible-Infrared Person Re-Identification},
 author={Lin, Xinyu and Li, Jinxing and Ma, Zeyu and Li, Huafeng and Li, Shuang and Xu, Kaixiong and Lu, Guangming and Zhang, David},
 booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
 pages={20973--20982},
 year={2022}
}