Home

Awesome

Official implementation for DINE

DINE: Domain Adaptation from Single and Multiple Black-box Predictors

Problem setting (Black-box UDA):

<img src="figs/problem.png" width="500" div align=center/>

Our framework (Distill-and-Finetune):

<img src="figs/method.png" width="800" div align=center/>

Prerequisites:

Demo:

Results:

<img src="figs/result-home.png" width="800" div align=center/>

Citation

If you find this code useful for your research, please cite our paper

@inproceedings{liang2022dine,
    title={DINE: Domain Adaptation from Single and Multiple Black-box Predictors},
    author={Liang, Jian and Hu, Dapeng and Feng, Jiashi and He, Ran},
    booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
    year={2022}
}

Contact