• Search by Module
• Search by Words
• Search Projects

• Java
• Python
• JavaScript
• TypeScript
• C++
• Scala
• Blog

Project: Deep-Person (GitHub Link)

• Deep-Person-master
  • environment.yml
  • examples
    • deep.py
  • LICENSE
  • README.md
  • reid
    • datasets
      • dukemtmc.py
      • viper.py
      • cuhk03_label.py
      • __init__.py
      • market1501.py
      • cuhk01.py
      • cuhk03_detect.py
    • utils
      • meters.py
      • osutils.py
      • serialization.py
      • data
        • sampler.py
        • preprocessor.py
        • __init__.py
        • dataset.py
        • transforms.py
      • __init__.py
      • logging.py
    • evaluation_metrics
      • ranking.py
      • classification.py
      • __init__.py
    • evaluators.py
    • trainers.py
    • models
      • resnet.py
      • deep_model.py
      • __init__.py
      • inception.py
    • dist_metric.py
    • feature_extraction
      • database.py
      • cnn.py
      • __init__.py
    • __init__.py
    • metric_learning
      • kissme.py
      • euclidean.py
      • __init__.py
    • loss
      • triplet.py
      • __init__.py
      • deep_loss.py
      • oim.py
  • .gitignore

Deep-Person: Learning Discriminative Deep Features for Person Re-Identification

Code for Deep-Person: Learning Discriminative Deep Features for Person Re-Identification.

Prerequisites

Deep-Person is developed and tested with Pytorch 0.2 and Python 3.6.

Anaconda is required to finish the Installation.

Installation

cd DeepPerson/
conda env create -f environment.yml

This will create an environment named "deepperson". (use conda list to see all environments)

Datasets Preparation

Market1501

DukeMTMC-Reid

CUHK03

Training and evaluation

NOTE: You must activate "deepperson" environment first before running the code.

To activate "deepperson" environment:

conda activate deepperson

To train a model:

cd DeepPerson/
python examples/deep.py -d market1501 --logs-dir logs/market

To evaluate a pretrained model:

cd DeepPerson/
python examples/deep.py -d market1501 --resume logs/market/checkpoint.pth.tar --evaluate

We provide a pretrained model on Market1501 which can be found at our release page.

Citation

If you find this project helpful for your research, please cite the following paper:

@article{xbai2017deepperson,
  author = {Xiang Bai and
               Mingkun Yang and
               Tengteng Huang and
               Zhiyong Dou and
               Rui Yu and
               Yongchao Xu},
  title   = {Deep-Person: Learning Discriminative Deep Features for Person Re-Identification},
  journal = {arXiv preprint arXiv:1711.10658},
  year    = {2017},
}

IMPORTANT NOTICE: Although this software is licensed under MIT, our intention is to make it free for academic research purposes. If you are going to use it in a product, we suggest you contact us regarding possible patent issues.

Acknowledgements

The code is based on open-reid. We sincerely thank for the great work.