https://github.com/TuSimple/rl-multishot-reid

Multi-shot Pedestrian Re-identification via Sequential Decision Making (CVPR2018)
https://github.com/TuSimple/rl-multishot-reid

Last synced: 2 months ago
JSON representation

Multi-shot Pedestrian Re-identification via Sequential Decision Making (CVPR2018)

• Host: GitHub
• URL: https://github.com/TuSimple/rl-multishot-reid
• Owner: TuSimple
• Created: 2017-12-21T02:57:36.000Z (about 7 years ago)
• Default Branch: master
• Last Pushed: 2017-12-21T02:58:35.000Z (about 7 years ago)
• Last Synced: 2024-08-01T22:41:04.515Z (5 months ago)
• Language: Python
• Homepage:
• Size: 51.8 KB
• Stars: 93
• Watchers: 16
• Forks: 26
• Open Issues: 5
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• Awesome-MXNet - rl-multishot-reid

README

        
* [Multi-shot Re-identification](#1)

* [Preparations](#1.1)

* [Usage](#1.2)

Multi-shot Re-identification Based on Reinforcement Learning


---

Training and testing codes for multi-shot Re-Identification. Currently, these codes are tested on the PRID-2011 dataset, iLiDS-VID dataset and MARS dataset. For algorithm details and experiment results, please refer our paper: [Multi-shot Pedestrian Re-identification via Sequential Decision Making](https://arxiv.org/abs/1712.07257)

Preparations


---

Before starting running this code, you should make the following preparations:

* Download the [MARS](http://www.liangzheng.com.cn/Project/project_mars.html)

, [iLIDS-VID](http://www.eecs.qmul.ac.uk/~xiatian/downloads_qmul_iLIDS-VID_ReID_dataset.html) and [PRID-2011](https://www.tugraz.at/institute/icg/research/team-bischof/lrs/downloads/PRID11/).

* Install MXNet following the [instructions](http://mxnet.io/get_started/index.html#setup-and-installation) and install the python interface. Currently the repo is tested on commit e06c55.

Usage


---

* Download the datasets and unzip.

* Prepare data file. Generate image list file according to the file `preprocess_ilds_image.py`

, `preprocess_prid_image.py` and `preprocess_mars_image.py` under `baseline` folder.

* The code is split to two stage, the first stage is a image based re-id task,

please refer the script `run.sh` in `baseline` folder. The codes for this stage is based on [this repo](https://github.com/TuSimple/re-identification). The usage is:

```shell

sh run.sh $gpu $dataset $network $recfloder

```

e.g. If you want to train MARS dataset on gpu 0 using inception-bn, please run:

```shell

sh run.sh 0 MARS inception-bn /data3/matt/MARS/recs

```

* The second stage is a multi-shot re-id task based on reinforcement learning.

Please refer the script `run.sh` in `RL` folder. The usage is:

```shell

sh run.sh $gpu $unsure-penalty $dataset $network $recfloder

```

* For evaluation, please use `baseline/baseline_test.py` and `RL/find_eg.py`. In `RL/find_eg.py`, we also show some example episodes with good quality generated by our algorithm.