JSIA-ReID

This is the official implementation for JSIA-ReID. Please refer our paper for more details:

[Paper] [Neural Network 2020] Cross-Modality Paired-Images Generation and Augmentation for RGB-Infrared Person Re-Identification

[Paper, Poster] [AAAI2020] Cross-Modality Paired-Images Generation for RGB-Infrared Person Re-Identification

Bibtex

If you find the code useful, please consider citing our paper:

@InProceedings{wang2020cross,
  title={Cross-modality paired-images generation and augmentation for RGB-infrared person re-identification},
  author={Guan'an, Wang and Yang, Yang and Zhang, Tianzhu and Cheng, Jian and Hou, Zengguang and Tiwari, Prayag and Pandey, Hari Mohan and others},
  journal={Neural Networks},
  volume={128},
  pages={294--304},
  year={2020},
  publisher={Elsevier}
}

@InProceedings{wang2020crossmodality,
author = "Guan-An {Wang} and Tianzhu {Zhang} and Yang {Yang} and Jian {Cheng} and Jianlong {Chang} and Xu {Liang} and Zengguang {Hou}",
title = {Cross-Modality Paired-Images Generation for RGB-Infrared Person Re-Identification},
booktitle = {AAAI-20 AAAI Conference on Artificial Intelligence},
year = {2020}
}

Dependencies

• Anaconda (Python 3.7)
• PyTorch 1.1.0
• GPU Memory >= 20G
• Memory >= 20G

Dataset Preparation

• SYSU-MM01 Dataset [link]

Train

# train, please replace sysu-mm01-path with your own path
python main.py --dataset_path sysu-mm01-path --mode train

Test with Pre-trained Model

• pretrained model (Google Drive， Baidu Disk(pwd:656y)), please download all the 4 files into a folder.
• test with the pre-trained model

# test with pretrained model, please replace sysu-mm01-path and pretrained-model-path with your own paths
python main.py --dataset_path sysu-mm01-path --mode test --pretrained_model_path pretrained-model-path --pretrained_model_epoch 649

Experimental Results

• Settings: We trained our model with 2 GTX1080ti GPUs.

• Comparison with SOTA

License

This repo is released under the MIT License.

Contacts

If you have any question about the project, please feel free to contact with me.

E-mail: [email protected]