This is the implementation of Maximum Classifier Discrepancy for digits classification and semantic segmentation in Pytorch. The code is written by Kuniaki Saito. The work was accepted by CVPR 2018 Oral.
Maximum Classifier Discrepancy for Domain Adaptation: [Project][Paper (arxiv)].
Go to classification or segmentation folder and see the instruction for each task.
If you use this code for your research, please cite our papers (This will be updated when cvpr paper is publicized).
@article{saito2017maximum,
title={Maximum Classifier Discrepancy for Unsupervised Domain Adaptation},
author={Saito, Kuniaki and Watanabe, Kohei and Ushiku, Yoshitaka and Harada, Tatsuya},
journal={arXiv preprint arXiv:1712.02560},
year={2017}
}