Unsupervised Cross-Domain Image Retrieval via Prototypical Optimal Transport

Code release for Unsupervised Cross-Domain Image Retrieval via Prototypical Optimal Transport [accepted by AAAI 2024]

Requirements

• Python 3.8.8
• Pytorch 1.8.0
• numpy 1.23.5
• scikit-learn 1.2.2
• faiss-gpu 1.7.2

Datasets

• Office-Home Dataset
• DomainNet Dataset

Pre-trained model

• MoCo v2 model: Download the MoCo v2 model trained after 800 epochs.

Run code

For DomainNet:

CUDA_VISIBLE_DEVICES=0 ./Scripts/DomainNet.sh

For Office-Home:

CUDA_VISIBLE_DEVICES=0 ./Scripts/Office-Home.sh

Model checkpoints

Our trained models can be downloaded as following.

• checkpoints

Acknowledgement

This repository is built based on the source code for Feature Representation Learning for Unsupervised Cross-domain Image Retrieval

Cite our work

If you find this repository useful in your research, please consider citing:

@article{li2024unsupervised,
  title={Unsupervised Cross-Domain Image Retrieval via Prototypical Optimal Transport},
  author={Li, Bin and Shi, Ye and Yu, Qian and Wang, Jingya},
  journal={arXiv preprint arXiv:2402.18411},
  year={2024}
}