ZUNIT

Pytorch implementation of our paper: "Zero-shot unsupervised image-to-image translation via exploiting semantic attributes".

<p align="center"> <img src='images/framework.png' align="center" width='90%'> </p>

Dependencies

you can install all the dependencies by

pip install -r requirements.txt

Datasets

• Download CUB dataset.
• Unzip the birds.zip at ./dataset.

Training

• To view training results and loss plots, run

python -m visdom.server -p 8080

and click the URL http://localhost:8080.

• Run

bash ./scripts/train_bird.sh

Testing

• Run

bash ./scripts/test_bird.sh

• The testing results will be saved in checkpoints/{exp_name}/results directory.

Results

<p align="center"> <img src='images/results1.png' align="center" width='90%'> </p> <p align="center"> <img src='images/results2.png' align="center" width='90%'> </p>

Bibtex

If this work is useful for your research, please consider citing :

@article{CHEN2022104489,
title = {Zero-shot unsupervised image-to-image translation via exploiting semantic attributes},
journal = {Image and Vision Computing},
pages = {104489},
year = {2022},
issn = {0262-8856},
doi = {https://doi.org/10.1016/j.imavis.2022.104489},
author = {Yuanqi Chen and Xiaoming Yu and Shan Liu and Wei Gao and Ge Li}
}

Acknowledgement

The code used in this research is inspired by DMIT and FUNIT.

Contact

Feel free to contact me if there is any questions (cyq373@pku.edu.cn).