This repository provides the official PyTorch implementation of our paper "Painting from Part".

Our paper can be found in https://openaccess.thecvf.com/content/ICCV2021/papers/Guo_Painting_From_Part_ICCV_2021_paper.pdf.

• Linux
• Python 3.7
• NVIDIA GPU + CUDA CuDNN

• Clone this repo:

git clone https://github.com/zhenglab/partpainting.git
cd partpainting

• Install PyTorch and 1.7 and other dependencies (e.g., torchvision).
  • For Conda users, you can create a new Conda environment using conda create --name <env> --file requirements.txt.

Please change the pathes to your dataset path in datasets folder.

The code defaults to regular outpainting task, and you may change mask types for other purpose in src/dataset.py and src/utils.py.

python train.py --path=$configpath$

For example: python train.py --path=./checkpoints/celeba-hq/

The model is automatically saved every 50,000 iterations, please rename the file g.pth_$iter_number$ to g.pth and then run testing command.

python test.py --path=$configpath$ 

For example: python test.py --path=./checkpoints/celeba-hq/

Irregular outpainting mask can be obtained from https://pan.baidu.com/s/1cyL5Cp0OddyNggZPmMvl8A?pwd=jk2m. Extraction code: jk2m.

@inproceedings{guo@painting,
author = {Guo, Dongsheng and Zhao, Haoru and Cheng, Yunhao and Zheng, Haiyong and Gu, Zhaorui and Zheng, Bing},
title = {Painting from Part},
booktitle = {ICCV},
year = {2021}
}