Unpaired Caricature Generation with Multiple Exaggerations (TMM 2021)
The official pytorch implementation of our TMM paper CariMe: Unpaired Caricature Generation with Multiple Exaggerations.
CariMe: Unpaired Caricature Generation with Multiple Exaggerations
Zheng Gu, Chuanqi Dong, Jing Huo, Wenbin Li, and Yang Gao
Paper: https://ieeexplore.ieee.org/abstract/document/9454341/
data
folder and align the dataset by running the following script:python alignment.py
Train the Warper:
python train_warper.py
Train the Styler:
python train_styler.py
python test_warper.py --scale 1.0
python test_styler.py
python main_generate.py --model_path_warper pretrained/warper.pt --model_path_styler pretrained/styler.pt
python main_generate_single_image.py --model_path_warper pretrained/warper.pt --model_path_styler pretrained/styler.pt --input_path images/Meg Ryan/P00015.jpg --generate_num 5 --scale 1.0
The above command will translate the input photo into 5 caricatures with different exaggerations and styles:
The pre-trained models are shared here.
If you use this code for your research, please cite our paper.
@article{gu2021carime,
title={CariMe: Unpaired Caricature Generation with Multiple Exaggerations},
author={Gu, Zheng and Dong, Chuanqi and Huo, Jing and Li, Wenbin and Gao, Yang},
journal={IEEE Transactions on Multimedia},
year={2021},
publisher={IEEE}
}