SymNets

Official PyTroch implementation for "Domain-Symnetric Networks for Adversarial Domain Adaptation (CVPR 2019)".

News!

An extension of this work is recently accepted by TPAMI 2020, including

A new theoretical framework closely supports/motivates a series of algorithms, including SymNets and MCD.
A algorithm improvement and an unified framework for adversarial UDA.
Excellent results of SymNets on tasks of partial and open set UDA.

TPAMI Paper "Unsupervised Multi-Class Domain Adaptation: Theory, Algorithms, and Practice". and Codes .

Prerequisites

Linux

NVIDIA GPU + CUDA (may CuDNN) and corresponding PyTorch framework (version 0.5.0)

Python 3.6

Training and Evaluation

Please refer to 'run.sh'

Citation

@inproceedings{zhang2019domain,
  title={Domain-symmetric networks for adversarial domain adaptation},
  author={Zhang, Yabin and Tang, Hui and Jia, Kui and Tan, Mingkui},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={5031--5040},
  year={2019}
}
@article{zhang2020unsupervised,
    title={Unsupervised multi-class domain adaptation: Theory, algorithms, and practice},
    author={Zhang, Yabin and Deng, Bin and Tang, Hui and Zhang, Lei and Jia, Kui},
    journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
    year={2020},
    publisher={IEEE}
}

Contact

If you have any problem about our code, feel free to contact

zhang.yabin@mail.scut.edu.cn

or describe your problem in Issues.

abutaufique/SymNets