SOLIDER on [Person Search]

This repo provides details about how to use SOLIDER pretrained representation on person search task. We modify the code from SeqNet, and you can refer to the original repo for more details.

Installation and Datasets

Details of installation and dataset preparation can be found in SeqNet.

Prepare Pre-trained Models

You can download models from SOLIDER, or use SOLIDER to train your own models. Before training, you should convert the models first.

python convert_model.py path/to/SOLIDER/log/lup/swin_tiny/checkpoint.pth path/to/SOLIDER/log/lup/swin_tiny/checkpoint_tea.pth

Training

We utilize 1 GPU for training. Please modify the ckpt and OUTPUT_DIR in the bash file.

sh run.sh

Performance

Method	Model	CUHK-SYSU (mAP/R1)	PRW (mAP/R1)
SOLIDER	Swin Tiny	94.91/95.72	56.84/86.78
SOLIDER	Swin Small	95.46/95.79	59.84/86.73
SOLIDER	Swin Base	94.93/95.52	59.72/86.83

• We use the pretrained models from SOLIDER.
• The semantic weight is set to 0.6 in these experiments.

Citation

If you find this code useful for your research, please cite our paper

@inproceedings{chen2023beyond,
  title={Beyond Appearance: a Semantic Controllable Self-Supervised Learning Framework for Human-Centric Visual Tasks},
  author={Weihua Chen and Xianzhe Xu and Jian Jia and Hao Luo and Yaohua Wang and Fan Wang and Rong Jin and Xiuyu Sun},
  booktitle={The IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2023},
}