Cross-Domain Adaptive Clustering for Semi-Supervised Domain Adaptation

This is a Pytorch implementation of "Cross-Domain Adaptive Clustering for Semi-Supervised Domain Adaptation" accepted by CVPR2021. More details of this work can be found in our paper: [Arxiv] or [OpenAccess].

Our code is based on SSDA_MME implementation.

Install

pip install -r requirements.txt

The code is written in Python 3.8.5, but should work for other versions with some modifications.

Data preparation

Refer to SSDA_MME and our paper.

Training

(1) To run training on DomainNet in the 3-shot scenario using alexnet,

python main.py --dataset multi --source real --target sketch --net alexnet --num 3 --lr_f 1.0 --multi 0.1 --save_check

(2) To run training on Office-Home in the 3-shot scenario using alexnet,

python main.py --dataset office_home --source Real --target Art --net alexnet --num 3 --lr_f 1.0 --multi 0.1 --steps 20000 --save_check

(3) To run training on Office31 in the 3-shot scenario using alexnet,

python main.py --dataset office --source webcam --target amazon --net alexnet --num 3 --lr_f 1.0 --multi 0.1 --steps 5000 --save_check

Citation

If you consider using this code or its derivatives, please consider citing:

@InProceedings{li2021cross,
    author    = {Li, Jichang and Li, Guanbin and Shi, Yemin and Yu, Yizhou},
    title     = {Cross-Domain Adaptive Clustering for Semi-Supervised Domain Adaptation},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2021},
    pages     = {2505-2514}
}

Contact

Please feel free to contact the first author, namely Li Jichang, with an Email address [email protected], if you have any questions.