Feature Erasing and Diffusion Network for Occluded Person Re-Identification (CVPR2022)
Pytorch implementation for the occluded person reid algorithm described in the paper Feature Erasing and Diffusion Network for Occluded Person Re-Identification (CVPR2022)
Pipline
Experiment Results on Holistic and Occluded Person ReID Datasets
TransReID
Retrieve Comparison betweenRequirements
Installation
Please refer to TransReID
Dataset Preparation
Please download Occluded-Duke dataset and cropped patches. Meanwhile place cropped patches into Occluded-Duke (just because of dataloader).
Pretrained Model Preparison
Please download pretrained ViT backbone in advance.
Model training and testing
before training and testing, please update config file accordingly. Around 13G GPU memory is required.
python train.py
Citation
If you find this code useful for your research, please cite our paper
@inproceedings{wang2022feature,
title={Feature Erasing and Diffusion Network for Occluded Person Re-Identification},
author={Wang, Zhikang and Zhu, Feng and Tang, Shixiang and Zhao, Rui and He, Lihuo and Song, Jiangning},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={4754--4763},
year={2022}
}
Contact
If you have any question, please feel free to contact us. E-mail: zhikang.wang@monash.edu