ADSPM: Attribute-Driven Spontaneous Motion in Unpaired Image Translation
This repository provides a PyTorch implementation of ADSPM.
Requirements
- Python 3.6
- PyTorch 0.4.1+
Usage
- Download pretrained model.
We provide 'Smiling' model in 256x256 and 512x512 resolution. [Download]
- Testing with pretrained model with sample images:
python test_samples.py --ckpt PRETRAINED_MODEL_PATH # 256x256 resolution model
python test_samples512.py --ckpt PRETRAINED_MODEL_PATH # 512x512 resolution model
Then the generated images will be saved in output
directory.
Citation
If you use this code for your research, please cite our paper.
@inproceedings{wu2019attribute,
title={Attribute-Driven Spontaneous Motion in Unpaired Image Translation},
author={Wu, Ruizheng and Tao, Xin and Gu, Xiaodong and Shen, Xiaoyong and Jia, Jiaya},
booktitle={Proceedings of the IEEE International Conference on Computer Vision (ICCV)},
year={2019}
}