wangxu-scu / CoDA

PyTorch implementation for Correspondence-Free Domain Alignment for Unsupervised Cross-Domain Image Retrieval (AAAI 2023)

CoDA

PyTorch implementation for Correspondence-Free Domain Alignment for Unsupervised Cross-Domain Image Retrieval (AAAI 2023).

Introduction

CoDA framework

Requirements

• Python 3.8
• PyTorch (1.10.0)
• numpy
• scikit-learn
• faiss (1.7.2)

Datasets

The directory structure of datasets.

datasets
├── OfficeHomeDataset_10072016 # 
│     ├── Art
│     ├── Clipart
│     ├── Product
│     ├── Real_World
│     ├── ......

Training and Evaluation

cd shells
sh run_coda.sh

Citation

If CoDA is useful for your research, please consider citing the paper:

@inproceedings{Wang2023CoDA,
    author = {Wang, Xu and Peng, Dezhong and Yan, Ming and Hu, Peng},
    title = {Correspondence-Free Domain Alignment for Unsupervised Cross-Domain Image Retrieval},
    year = {2023},
    booktitle = {The Thirty-Seventh AAAI Conference on Artificial Intelligence},
    series = {AAAI 2023}
}

License

Apache License 2.0