Skip to content
/ DMML Public

code for ICCV19 paper "Deep Meta Metric Learning"

Notifications You must be signed in to change notification settings

CHENGY12/DMML

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

40 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

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.
  • DukeMTMC-reID dataset can be downloaded from here.

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.

@inproceedings{chen2019deep,
  title={Deep Meta Metric Learning},
  author={Chen, Guangyi and Zhang, Tianren and Lu, Jiwen and Zhou, Jie},
  booktitle={ICCV},
  year={2019}
}