P2C

Code for our paper "Unpaired Photo-to-Caricature Translation on Faces in the Wild" (arXiv:1711.10735).

Citation

@article{zheng2019unpaired,
  title={Unpaired Photo-to-Caricature Translation on Faces in the Wild},
  author={Zheng, Ziqiang and Chao, Wang and Yu, Zhibin and Wang, Nan and Zheng, Haiyong and Zheng, Bing},
  journal={Neurocomputing},
  doi={10.1016/j.neucom.2019.04.032},
  year={2019}
}

Installation

1. We use Miniconda3 for the basic environment. If you installed the Miniconda3 in path Conda_Path, please install tensorflow-gpu using the command Conda_Path/bin/conda install -c anaconda tensorflow-gpu==1.8.
2. Install dependencies by Conda_Path/bin/pip install -r requirements.txt (if necessary). The requirements.txt file is provided in this package.

Data preparation

├── datasets
   └── demo
       ├── trainA
           ├── 000001.jpg (The traint image that you want, name does not matter)
           ├── 000002.jpg
           └── ...
       ├── trainB
           ├── 000001.jpg (The traint image that you want, name does not matter)
           ├── 000002.jpg
           └── ...
       ├── testA
           ├── a.jpg (The test image that you want)
           ├── b.png
           └── ...
       ├── testB
           ├── a.jpg (The test image that you want)
           ├── b.png
           └── ...

usage

• base.py: train and test model of P2C.
• utils.py: basic utils of P2C.

train

Conda_Path/bin/python base.py --phase train --dataset_dir demo --checkpoint ./checkpoint ./checkpoints/demo --sample_dir ./checkpoints/demo/sample --epoch 120 --gpu 0

test

Conda_Path/bin/python base.py --phase test --dataset_dir demo --checkpoint ./checkpoint ./checkpoints/demo --test_dir ./checkpoints/demo/test --gpu 0