https://github.com/CHENGY12/DMML

code for ICCV19 paper "Deep Meta Metric Learning"
https://github.com/CHENGY12/DMML

baseline meta-learning metric-learning person-reidentification pytorch resnet-50 vehicle-reidentification

Last synced: 13 days ago
JSON representation

code for ICCV19 paper "Deep Meta Metric Learning"

• Host: GitHub
• URL: https://github.com/CHENGY12/DMML
• Owner: CHENGY12
• Created: 2019-07-24T14:32:27.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2024-07-25T10:12:44.000Z (4 months ago)
• Last Synced: 2024-07-31T23:45:10.701Z (3 months ago)
• Topics: baseline, meta-learning, metric-learning, person-reidentification, pytorch, resnet-50, vehicle-reidentification
• Language: Python
• Size: 37.1 KB
• Stars: 108
• Watchers: 3
• Forks: 18
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesome-few-shot-meta-learning - code - official (PyTorch)

README

        
# Deep Meta Metric Learning (DMML)

This repo contains PyTorch code for ICCV19' paper: Deep Meta Metric Learning, including person re-identification experiments on Market-1501 and DukeMTMC-reID datasets.

## Requirements

- Python 3.6+

- PyTorch 0.4

- tensorboardX 1.6

To install all python packages, please run the following command:

```

pip install -r requirements.txt

```

## Datasets

### Downloading

- Market-1501 dataset can be downloaded from [here](http://www.liangzheng.org/Project/project_reid.html).

- DukeMTMC-reID dataset can be downloaded from [here](http://vision.cs.duke.edu/DukeMTMC/).

### Preparation

After downloading the datasets above, move them to the `datasets/` folder in the project root directory, and rename dataset folders to 'market1501' and 'duke' respectively. I.e., the `datasets/` folder should be organized as:

```

|-- market1501

    |-- bounding_box_train

    |-- bounding_box_test

    |-- ...

|-- duke

    |-- bounding_box_train

    |-- bounding_box_test

    |-- ...

```

## Usage

### Training

After adding dataset directory in `demo.sh`, simply run the following command to train DMML on Market-1501:

```

bash demo.sh

```

Usage instructions of all training parameters can be found in `config.py`.

### Evaluation

To evaluate the performance of a trained model, run

```

python eval.py

```

which will output Rank-1, Rank-5, Rank-10 and mAP scores.

### Citation

Please use the citation provided below if it is useful to your research:

Guangyi Chen, Tianren Zhang, Jiwen Lu, and Jie Zhou, Deep Meta Metric Learning, ICCV, 2019.

```bash

@inproceedings{chen2019deep,

  title={Deep Meta Metric Learning},

  author={Chen, Guangyi and Zhang, Tianren and Lu, Jiwen and Zhou, Jie},

  booktitle={ICCV},

  year={2019}

}

```