One Shot Domain Adaptation For FaceGeneration

This repository contains our PyTorch evaluation code and pretrained models for One Shot Domain Adaptation For FaceGeneration.

Given a single target image, we could reconstruct the image using StyleGAN and generate many more similar looking images.

For more details, please see One-Shot Domain Adaptation For Face Generation by Harry (Chao) Yang and Ser-Nam Lim.

If you find this repo useful in your research, please consider citing our work:

@InProceedings{Yang_2020_CVPR,
author = {Yang, Chao and Lim, Ser-Nam},
title = {One-Shot Domain Adaptation for Face Generation},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2020}
}

Run

Please download StyleGAN pretrained pytorch models and dlib landmark detector model from Google Drive

python one_shot_domain_adaptation.py --img_name ./test_data/autoencoder.png --verbose

All intermediate and final results are saved under test_results/test.

License

One Shot Domain Adaptation for Face Generation is CC-BY-4.0 licensed, as found in the LICENSE file. Some of the codes are adapted from NVIDIA FFHQ and StyleGAN PyTorch and share their licenses.

Contributing

We actively welcome your pull requests! Please see CONTRIBUTING and CODE_OF_CONDUCT for more info.