Demo code for Cascaded SR-GAN for Scale-Adaptive Low Resolution Person Re-identification in IJCAI-18. We use SALR-VIPeR as an example.
Download the VGG model to current folder. Download the Basic Re-id model, pre-trained on Market-1501, to current folder. Download the dataset to
.\dataset\
.
Run
python train_csr_gan_viper.py
Run
python get_feature_viper.py
Please cite this paper in your publications if it helps your research:
@inproceedings{wang2018cascaded,
title={Cascaded SR-GAN for Scale-Adaptive Low Resolution Person Re-identification.},
author={Wang, Zheng and Ye, Mang and Yang, Fan and Bai, Xiang and Satoh, Shin'ichi},
booktitle={IJCAI},
year={2018}
}
If you also use the SALR datasets, please kindly cite:
@inproceedings{wang2016scale,
title={Scale-Adaptive Low-Resolution Person Re-Identification via Learning a Discriminating Surface.},
author={Wang, Zheng and Hu, Ruimin and Yu, Yi and Jiang, Junjun and Liang, Chao and Wang, Jinqiao},
booktitle={IJCAI},
pages={2669--2675},
year={2016}
}
Please also kindly cite:
@article{ledig2017photo,
title={Photo-realistic single image super-resolution using a generative adversarial network},
author={Ledig, Christian and Theis, Lucas and Husz{\'a}r, Ferenc and Caballero, Jose and Cunningham, Andrew and Acosta, Alejandro and Aitken, Andrew and Tejani, Alykhan and Totz, Johannes and Wang, Zehan and others},
journal={arXiv preprint},
year={2017}
}
This code helps us very much. Link: https://github.com/tensorlayer/srgan
Contact: [email protected]