Code release for Unsupervised Cross-Domain Image Retrieval via Prototypical Optimal Transport [accepted by AAAI 2024]
- Python 3.8.8
- Pytorch 1.8.0
- numpy 1.23.5
- scikit-learn 1.2.2
- faiss-gpu 1.7.2
- MoCo v2 model: Download the MoCo v2 model trained after 800 epochs.
For DomainNet:
CUDA_VISIBLE_DEVICES=0 ./Scripts/DomainNet.sh
For Office-Home:
CUDA_VISIBLE_DEVICES=0 ./Scripts/Office-Home.sh
Our trained models can be downloaded as following.
This repository is built based on the source code for Feature Representation Learning for Unsupervised Cross-domain Image Retrieval
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}
}