KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#dsh #dsh-plugin #ai-agent #developer-tools #codex #typescript #cli #deepseek-harness #claude-code #ai-agents #mcp #rust

2016_person_re-ID

★ 264 Open GitHub ↗

TOMM2017 A Discriminatively Learned CNN Embedding for Person Re-identification

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

## A Discriminatively Learned CNN Embedding for Person Re-identification

![](https://github.com/layumi/2016_person_re-ID/blob/master/paper.jpg)

In this package, we provide our training and testing code written in [Matconvnet](http://www.vlfeat.org/matconvnet/) for the paper [A Discriminatively Learned CNN Embedding for Person Re-identification](https://arxiv.org/abs/1611.05666).
 
We also include matconvnet-beta23 which has been modified for our paper. All codes have been test on Ubuntu14.04 and Ubuntu16.04 with Matlab R2015b.

**This code is ONLY released for academic use.**

## Resources

* [Weihang Chen](https://github.com/ahangchen) also realizes our paper in [Keras](https://github.com/ahangchen/rank-reid/tree/release). ([![GitHub stars](https://img.shields.io/github/stars/ahangchen/rank-reid.svg?style=flat&label=Star)](https://github.com/ahangchen/rank-reid/tree/release))
* [Xuanyi Dong](https://github.com/D-X-Y) also realizes our paper in [Caffe](https://github.com/D-X-Y/caffe-reid). ([![GitHub stars](https://img.shields.io/github/stars/D-X-Y/caffe-reid.svg?style=flat&label=Star)](https://github.com/D-X-Y/caffe-reid))
* [Zhun Zhong](https://github.com/zhunzhong07/IDE-baseline-Market-1501) provides a extensive [Caffe baseline code](https://github.com/zhunzhong07/IDE-baseline-Market-1501). You may check it. ([![GitHub stars](https://img.shields.io/github/stars/zhunzhong07/IDE-baseline-Market-1501.svg?style=flat&label=Star)](https://github.com/zhunzhong07/IDE-baseline-Market-1501))
* [Zhedong Zheng](https://github.com/layumi) provides a [strong Pytorch baseline](https://github.com/layumi/Person_reID_baseline_pytorch) ([![GitHub stars](https://img.shields.io/github/stars/layumi/Person_reID_baseline_pytorch.svg?style=flat&label=Star)](https://github.com/layumi/Person_reID_baseline_pytorch)) and realizes our paper in [Pytorch](https://github.com/layumi/Person-reID-verification). ([![GitHub stars](https://img.shields.io/github/stars/layumi/Person-reID-verification.svg?style=flat&label=Star)](https://github.com/layumi/Person-reID-verification))

~What's new: We add the data preparation and evaluation codes for CUHK03.

~What's new: We make the code of model structure more easy to follow.

~What's new: We provide a better code for extract feature.

~What's new: We provide a faster evaluation code for Market-1501.

## Installation
1. Clone this repo
 
	```Shell
	git clone https://github.com/layumi/2016_person_re-ID.git
	cd 2016_person_re-ID
	mkdir data
	```

2. Download the pretrained model. 
	
	**This model is ONLY released for academic use.**
	You can find the pretrained model in [GoogleDriver](https://drive.google.com/open?id=0B0VOCNYh8HeRWks0V24xTlpKWkU) or [BaiduYun] (https://pan.baidu.com/s/1miG2OpM). Download and put them into the `./data`.

BaiduYun sometime changes the link. If you find the url fail, you can contact me to update it.
	
3. Compile matconvnet 
**(Note that I have included my Matconvnet in this repo, so you do not need to download it again. I has changed some codes comparing with the original version. For example, one of the difference is in `/matlab/+dagnn/@DagNN/initParams.m`. If one layer has params, I will not initialize it again, especially for pretrained model.)**

You just need to uncomment and modify some lines in `gpu_compile.m` and run it in Matlab. Try it~
	(The code does not support cudnn 6.0. You may just turn off the Enablecudnn or try cudnn5.1)

If you fail in compilation, you may refer to http://www.vlfeat.org/matconvnet/install/

## Dataset
* Download [Market1501 Dataset](http://www.liangzheng.org/Project/project_reid.html). [[Google]](https://drive.google.com/file/d/0B8-rUzbwVRk0c054eEozWG9COHM/view) [[Baidu]](https://pan.baidu.com/s/1ntIi2Op) The photos are taken in Tsinghua University.

* DukeMTMC-reID is a larger dataset in the same format of Market1501. The photos are taken in Duke University.
You can download it from [DukeMTMC-reID Dataset](https://github.com/layumi/DukeMTMC-reID_evaluation). We also upload the result to DukeMTMC-reID leaderboard.

* If you want to rehearsal our result on [CUHK03 Dataset](http://www.ee.cuhk.edu.hk/~rzhao/), you can simply change the number of kernel from 751 to 1367 in `resnet52_market.m` and recreate `net.mat`. Because there are 751 IDs in Market-1501 while 1367 training identities are in CUHK03. More information can be found in `cuhk03-prepare-eval` dir. We add the data preparation and evaluation codes for CUHK03.

* Training dataset for Oxford5k (http://cmp.felk.cvut.cz/cnnimageretrieval/)

## Test 
1. Run `test/test_gallery_query_crazy.m` to extract the features of images in the gallery and query set. They will store in a .mat file. Then you can use it to do evaluation.
2. Evaluate feature on the Market-1501. Run `evaluation/zzd_evaluation_res_faster.m`. You can get the following Single-query Result.

| Methods               | Rank@1 | mAP    | 
| --------              | -----  | ----   | 
| Ours* (SQ)            | 80.82% | 62.30% | 
| Ours* (MQ-avg)        | 86.67% | 70.16% | 
| Ours* (MQ-max)        | 86.76% | 70.68% | 
| Ours* (MQ-max+rerank) | 86.67% | 72.55% |

*Note that the result is slightly higher than the result reported in our paper.

*For multi-query result, you can use `evaluation/zzd_evaluation_res_fast.m` . It is slower than `evaluation/zzd_evaluation_res_faster.m`  since it need to extract extra features. (The evaluation code is modified from the [Market-1501 Baseline Code](http://www.liangzheng.org/Project/project_reid.html))

## FQA
1. What is multi-query setting?

Actually, we can get a sequence of the query under one camera instead of one image. Then we can use every image in this sequence to extract a query mean feature (mean of feature extracted from several images).
We call it multi-query. If we use this feature to do person retrieval, we usually get a better result.
But it use additional images (in 'Market-1501/gt_bboxes'). You can find more detail in the [original paper](http://www.cv-foundation.org/openaccess/content_iccv_2015/papers/Zheng_Scalable_Person_Re-Identification_ICCV_2015_paper.pdf).

## Train
1. Add your dataset path into `prepare_data.m` and run it. Make sure the code outputs the right image path.

2. Run `train_id_net_res_2stream.m` to have fun.

## 🔗 Related Projects

<p align="center"><i>Explore our ecosystem for Object Re-Identification 🔍</i></p>

<div align="center">
  <table>
    <tr>
      <td align="center" colspan="3">
        <a href="https://github.com/layumi/Person_reID_baseline_pytorch">
          <h3>⛹️</h3>
          <b>Person re-ID Baseline (PyTorch)</b>
        </a>
        <br><sub>A <b>Tiny, Friendly & Strong</b> PyTorch Baseline for Person / Vehicle Re-ID<br>with Hands-on Tutorial · The Community Standard since 2017</sub>
        <br><br>
        <a href="https://github.com/layumi/Person_reID_baseline_pytorch"><img src="https://img.shields.io/github/stars/layumi/Person_reID_baseline_pytorch.svg?style=social&label=Star" alt="GitHub stars"></a>
      </td>
    </tr>
    <tr>
      <td align="center" width="33%">
        <a href="https://github.com/layumi/3D-Magic-Mirror">
          <h3>🪞</h3>
          <b>3D Magic Mirror</b>
        </a>
        <br><sub>Clothing Reconstruction from a Single Image<br>via a Causal Perspective · npj AI'26</sub>
        <br><br>
        <a href="https://github.com/layumi/3D-Magic-Mirror"><img src="https://img.shields.io/github/stars/layumi/3D-Magic-Mirror.svg?style=social&label=Star" alt="GitHub stars"></a>
      </td>
      <td align="center" width="33%">
        <a href="https://github.com/NVlabs/DG-Net">
          <h3>✨</h3>
          <b>DG-Net</b>
        </a>
        <br><sub>Joint Generation + Re-ID Learning<br>CVPR'19 Oral · NVIDIA</sub>
        <br><br>
        <a href="https://github.com/NVlabs/DG-Net"><img src="https://img.shields.io/github/stars/NVlabs/DG-Net.svg?style=social&label=Star" alt="GitHub stars"></a>
      </td>
      <td align="center" width="33%">
        <a href="https://github.com/layumi/Person-reID_GAN">
          <h3>🎨</h3>
          <b>Person re-ID GAN</b>
        </a>
        <br><sub>GAN-based Augmentation (LSRO)<br>ICCV'17</sub>
        <br><br>
        <a href="https://github.com/layumi/Person-reID_GAN"><img src="https://img.shields.io/github/stars/layumi/Person-reID_GAN.svg?style=social&label=Star" alt="GitHub stars"></a>
      </td>
    </tr>
    <tr>
      <td align="center" width="33%">
        <a href="https://github.com/layumi/Image-Text-Embedding">
          <h3>📝</h3>
          <b>Language Person Search</b>
        </a>
        <br><sub>Text-based Person Retrieval<br>Dual-Path Embedding</sub>
        <br><br>
        <a href="https://github.com/layumi/Image-Text-Embedding"><img src="https://img.shields.io/github/stars/layumi/Image-Text-Embedding.svg?style=social&label=Star" alt="GitHub stars"></a>
      </td>
      <td align="center" width="33%">
        <a href="https://github.com/Shuyu-XJTU/APTM">
          <h3>🏷️</h3>
          <b>APTM</b>
        </a>
        <br><sub>Attribute Prompt Learning &amp; Text Matching<br>MALS Benchmark (1.5M pairs) · ACM MM'23</sub>
        <br><br>
        <a href="https://github.com/Shuyu-XJTU/APTM"><img src="https://img.shields.io/github/stars/Shuyu-XJTU/APTM.svg?style=social&label=Star" alt="GitHub stars"></a>
      </td>
      <td align="center" width="33%">
        <a href="https://github.com/Shuyu-XJTU/CMP">
          <h3>🚨</h3>
          <b>CMP</b>
        </a>
        <br><sub>Text-based Person <b>Anomaly</b> Search<br>PAB Benchmark (1M pairs) · ICCV'25 <b>Highlight</b></sub>
        <br><br>
        <a href="https://github.com/Shuyu-XJTU/CMP"><img src="https://img.shields.io/github/stars/Shuyu-XJTU/CMP.svg?style=social&label=Star" alt="GitHub stars"></a>
      </td>
    </tr>
    <tr>
      <td align="center" width="33%">
        <a href="https://github.com/layumi/person-reid-3d">
          <h3>🧊</h3>
          <b>3D Person re-ID</b>
        </a>
        <br><sub>Parameter-Efficient Re-ID<br>in the 3D Space (OG-Net)</sub>
        <br><br>
        <a href="https://github.com/layumi/person-reid-3d"><img src="https://img.shields.io/github/stars/layumi/person-reid-3d.svg?style=social&label=Star" alt="GitHub stars"></a>
      </td>
      <td align="center" width="33%">
        <a href="https://github.com/layumi/Pedestrian_Alignment">
          <h3>🚶</h3>
          <b>Pedestrian Alignment</b>
        </a>
        <br><sub>Pedestrian Alignment Network (PAN)<br>for Robust Re-ID</sub>
        <br><br>
        <a href="https://github.com/layumi/Pedestrian_Alignment"><img src="https://img.shields.io/github/stars/layumi/Pedestrian_Alignment.svg?style=social&label=Star" alt="GitHub stars"></a>
      </td>
      <td align="center" width="33%">
        <a href="https://github.com/layumi/2016_person_re-ID">
          <h3>🔁</h3>
          <b>2-Stream Person re-ID</b>
        </a>
        <br><sub>Verification + Identification<br>Baseline</sub>
        <br><br>
        <a href="https://github.com/layumi/2016_person_re-ID"><img src="https://img.shields.io/github/stars/layumi/2016_person_re-ID.svg?style=social&label=Star" alt="GitHub stars"></a>
      </td>
    </tr>
  </table>
</div>

---

<p align="center">
  ⭐ If you find our projects helpful, a <b>star</b> is the best support! ⭐
</p>

## Citation
Please cite this paper in your publications if it helps your research:
```bibtex
@article{zheng2016discriminatively,
  title={A Discriminatively Learned CNN Embedding for Person Re-identification},
  author={Zheng, Zhedong and Zheng, Liang and Yang, Yi},
  doi={10.1145/3159171},
  note={\mbox{doi}:\url{10.1145/3159171}},
  journal={ACM Transactions on Multimedia Computing Communications and Applications},
  year={2017}
}
```
## Acknowledge
Thanks for Xuanyi Dong to realize our paper in Caffe.

Thanks for Weihang Chen to realize our paper in Keras.

Thanks for Weihang Chen to report the bug in `prepare_data.m`.

## Related Repos
1. [Person re-ID with GAN](https://github.com/layumi/Person-reID_GAN)
2. [Pedestrian Alignment Network](https://github.com/layumi/Pedestrian_Alignment)

2016_person_re-ID

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README