Intermediary-guided Bidirectional Spatial-Temporal Aggregation Network for Video-based Visible-Infrared Person Re-Identification

Paper

Pipeline

Requirements

Installation

We use /torch >=1.8 / 24G RTX3090 for training and evaluation.

Prepare Datasets

mkdir data_original

mkdir data_anaglyph

There are many ways to generate anaglyph images, and you can also use the code (main_VCM.py) we provide.

Note that the organization, file name, and storage format of the original data and the anaglyph data should be consistent.

data
├── data_original
│   └── 
│   └── 
│   └── 
│   └── ..
├── data_anaglyph
│   └── 
│   └── 
│   └── 
│   └── ..

Training and Evaluation

python train.py

Later, we will upload our trained model(download), and you can load the model directly without training.

Contact

If you have any questions, please feel free to contact me. ( liuminghui_1997@163.com ).

Cite

@ARTICLE{10047982,
  author={Li, Huafeng and Liu, Minghui and Hu, Zhanxuan and Nie, Feiping and Yu, Zhengtao},
  journal={IEEE Transactions on Circuits and Systems for Video Technology}, 
  title={Intermediary-guided Bidirectional Spatial-Temporal Aggregation Network for Video-based Visible-Infrared Person Re-Identification}, 
  year={2023},
  volume={},
  number={},
  pages={1-1},
  doi={10.1109/TCSVT.2023.3246091}}

lhf12278/IBAN