## Introduction
This is the implementation for [Anchor-Free Person Search](https://arxiv.org/abs/2103.11617) in CVPR2021.
A brief introduction in Chinese can be found at https://zhuanlan.zhihu.com/p/359617800
## License
This project is released under the [Apache 2.0 license](LICENSE).
## Installation
This project is developed upon [MMdetection](https://github.com/open-mmlab/mmdetection), please refer to [install.md](docs/install.md) to install MMdetection.
We utilized mmcv=1.1.5, pytorch=1.7.0
## Dataset
Download [CUHK-SYSU](https://github.com/ShuangLI59/person_search) and [PRW](https://github.com/liangzheng06/PRW-baseline).
We provide coco-style annotation in [demo/anno](demo/anno).
For CUHK-SYSU, change the path of your dataset and the annotaion file in the [config file](configs/_base_/datasets/cuhk_detection_1000.py) L3, L38, L43, L48
For PRW, change the paths in these config files: [config1](configs/fcos/prw_base_focal_labelnorm_sub_ldcn_fg15_wd1-3.py) [config2](configs/fcos/prw_dcn_base_focal_labelnorm_sub_ldcn_fg15_wd7-4.py)
## Experiments
1. Train
```bash
sh run_train.sh
```
2. Test CUHK-SYSU
Change the paths in L59 and L72 in [test_results.py](tools/test_results.py)
```bash
sh run_test.sh
```
3. Test PRW
Change the paths in L127 and L128 in [test_results_prw.py](tools/test_results_prw.py)
```bash
sh run_test_prw.sh
```
## Performance
|Dataset|Model|mAP|Rank1| Config | Link |
|-----|-----|------|-----|------|-----|
|CUHK-SYSU|AlignPS| 93.1%|93.4%|[cfg](https://github.com/daodaofr/AlignPS/blob/master/configs/fcos/fcos_center-normbbox-centeronreg-giou_r50_caffe_fpn_gn-head_dcn_4x4_1x_cuhk_reid_1500_stage1_fpncat_dcn_epoch24_multiscale_focal_x4_bg-2_lconv3dcn_sub_triqueue_dcn0.py)| [model](https://drive.google.com/file/d/1WMvvxee15Enca_l9DYzCuOfP1f64zliy/view?usp=sharing)|
|CUHK-SYSU|AlignPS+|94.0%|94.5%|[cfg](https://github.com/daodaofr/AlignPS/blob/master/configs/fcos/fcos_center-normbbox-centeronreg-giou_r50_caffe_fpn_gn-head_dcn_4x4_1x_cuhk_reid_1500_stage1_fpncat_dcn_epoch24_multiscale_focal_x4_bg-2_lconv3dcn_sub_triqueue.py)| [model](https://drive.google.com/file/d/12AuG37IPkhyrpHG_kqpUzzoDEEkXlgne/view?usp=sharing)|
|PRW|AlignPS| 45.9%|81.9%|[cfg](https://github.com/daodaofr/AlignPS/blob/master/configs/fcos/prw_base_focal_labelnorm_sub_ldcn_fg15_wd1-3.py)| [model](https://drive.google.com/file/d/1QQNoYQTiO3FIiEpu0AtigGFIDf3wG2u5/view?usp=sharing)|
|PRW|AlignPS+|46.1%|82.1%|[cfg](https://github.com/daodaofr/AlignPS/blob/master/configs/fcos/prw_dcn_base_focal_labelnorm_sub_ldcn_fg15_wd7-4.py)| [model](https://drive.google.com/file/d/1O02EBrHglE1x-zk88QLLdXF-x6yebwBp/view?usp=sharing)|
## Citation
If you use this toolbox or benchmark in your research, please cite this project.
```
@inproceedings{yan2021alignps,
title={Anchor-Free Person Search},
author={Yichao Yan, Jinpeng Li, Jie Qin, Song Bai, Shengcai Liao, Li Liu, Fan Zhu, Ling Shao},
booktitle={CVPR},
year={2021}
}
```
Raw data
{
"_id": null,
"home_page": "https://github.com/open-mmlab/mmdetection",
"name": "mmps",
"maintainer": "",
"docs_url": null,
"requires_python": "",
"maintainer_email": "",
"keywords": "computer vision,object detection",
"author": "OpenMMLab",
"author_email": "openmmlab@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/29/4c/840bfd5f7813bba334a7c61d01cc3b6c60c73d651c24cad2b034edfc0764/mmps-2.4.0.tar.gz",
"platform": null,
"description": "## Introduction\n\nThis is the implementation for [Anchor-Free Person Search](https://arxiv.org/abs/2103.11617) in CVPR2021.\n\nA brief introduction in Chinese can be found at https://zhuanlan.zhihu.com/p/359617800\n\n\n\n\n## License\n\nThis project is released under the [Apache 2.0 license](LICENSE).\n\n\n## Installation\n\nThis project is developed upon [MMdetection](https://github.com/open-mmlab/mmdetection), please refer to [install.md](docs/install.md) to install MMdetection.\n\nWe utilized mmcv=1.1.5, pytorch=1.7.0\n\n\n## Dataset\n\nDownload [CUHK-SYSU](https://github.com/ShuangLI59/person_search) and [PRW](https://github.com/liangzheng06/PRW-baseline).\n\nWe provide coco-style annotation in [demo/anno](demo/anno).\n\nFor CUHK-SYSU, change the path of your dataset and the annotaion file in the [config file](configs/_base_/datasets/cuhk_detection_1000.py) L3, L38, L43, L48\n\nFor PRW, change the paths in these config files: [config1](configs/fcos/prw_base_focal_labelnorm_sub_ldcn_fg15_wd1-3.py) [config2](configs/fcos/prw_dcn_base_focal_labelnorm_sub_ldcn_fg15_wd7-4.py)\n\n\n\n## Experiments\n 1. Train\n\n ```bash\n sh run_train.sh\n ```\n 2. Test CUHK-SYSU\n\n Change the paths in L59 and L72 in [test_results.py](tools/test_results.py)\n\n ```bash\n sh run_test.sh\n ```\n 3. Test PRW\n\n Change the paths in L127 and L128 in [test_results_prw.py](tools/test_results_prw.py)\n\n ```bash\n sh run_test_prw.sh\n ```\n\n## Performance\n\n|Dataset|Model|mAP|Rank1| Config | Link |\n|-----|-----|------|-----|------|-----|\n|CUHK-SYSU|AlignPS| 93.1%|93.4%|[cfg](https://github.com/daodaofr/AlignPS/blob/master/configs/fcos/fcos_center-normbbox-centeronreg-giou_r50_caffe_fpn_gn-head_dcn_4x4_1x_cuhk_reid_1500_stage1_fpncat_dcn_epoch24_multiscale_focal_x4_bg-2_lconv3dcn_sub_triqueue_dcn0.py)| [model](https://drive.google.com/file/d/1WMvvxee15Enca_l9DYzCuOfP1f64zliy/view?usp=sharing)| \n|CUHK-SYSU|AlignPS+|94.0%|94.5%|[cfg](https://github.com/daodaofr/AlignPS/blob/master/configs/fcos/fcos_center-normbbox-centeronreg-giou_r50_caffe_fpn_gn-head_dcn_4x4_1x_cuhk_reid_1500_stage1_fpncat_dcn_epoch24_multiscale_focal_x4_bg-2_lconv3dcn_sub_triqueue.py)| [model](https://drive.google.com/file/d/12AuG37IPkhyrpHG_kqpUzzoDEEkXlgne/view?usp=sharing)| \n|PRW|AlignPS| 45.9%|81.9%|[cfg](https://github.com/daodaofr/AlignPS/blob/master/configs/fcos/prw_base_focal_labelnorm_sub_ldcn_fg15_wd1-3.py)| [model](https://drive.google.com/file/d/1QQNoYQTiO3FIiEpu0AtigGFIDf3wG2u5/view?usp=sharing)| \n|PRW|AlignPS+|46.1%|82.1%|[cfg](https://github.com/daodaofr/AlignPS/blob/master/configs/fcos/prw_dcn_base_focal_labelnorm_sub_ldcn_fg15_wd7-4.py)| [model](https://drive.google.com/file/d/1O02EBrHglE1x-zk88QLLdXF-x6yebwBp/view?usp=sharing)| \n\n\n## Citation\n\nIf you use this toolbox or benchmark in your research, please cite this project.\n\n```\n@inproceedings{yan2021alignps,\n title={Anchor-Free Person Search},\n author={Yichao Yan, Jinpeng Li, Jie Qin, Song Bai, Shengcai Liao, Li Liu, Fan Zhu, Ling Shao},\n booktitle={CVPR},\n year={2021}\n}\n```",
"bugtrack_url": null,
"license": "Apache License 2.0",
"summary": "OpenMMLab Detection Toolbox and Benchmark",
"version": "2.4.0",
"split_keywords": [
"computer vision",
"object detection"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "294c840bfd5f7813bba334a7c61d01cc3b6c60c73d651c24cad2b034edfc0764",
"md5": "19d89dc7bf8cfc05670587f5d0e40320",
"sha256": "76fcf09337a757d6366f8a4156a2a50ffac88afec424b4ca8d2dc506c85bee3e"
},
"downloads": -1,
"filename": "mmps-2.4.0.tar.gz",
"has_sig": false,
"md5_digest": "19d89dc7bf8cfc05670587f5d0e40320",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 476901,
"upload_time": "2023-02-11T14:16:17",
"upload_time_iso_8601": "2023-02-11T14:16:17.979162Z",
"url": "https://files.pythonhosted.org/packages/29/4c/840bfd5f7813bba334a7c61d01cc3b6c60c73d651c24cad2b034edfc0764/mmps-2.4.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-02-11 14:16:17",
"github": true,
"gitlab": false,
"bitbucket": false,
"github_user": "open-mmlab",
"github_project": "mmdetection",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"circle": true,
"requirements": [],
"lcname": "mmps"
}
JSON
