pytorch-MRGAN-master

Code of the moire removal method "Unsupervised Moire Pattern Removal for Recaptured Screen Images"

Results

Framework

Prerequisites

• Linux or macOS
• Python 3
• CPU or NVIDIA GPU + CUDA CuDNN

Getting Started

Installation

• Clone this repo:

git clone https://github.com/JerryLeolfl/pytorch-MRGAN-master.git
cd pytorch-MRGAN-master

• Install PyTorch and 0.4+ and other dependencies (e.g., torchvision, visdom and dominate).

train/test

• Get the whole dataset(key:8wm4).
• To view training results and loss plots, run python -m visdom.server and click the URL http://localhost:8097.
• Train a model:

python train.py --dataroot ./path/to/trainset --name moire_cyclegan --model cycle_gan

To see more intermediate results, check out ./checkpoints/moire_cyclegan/web/index.html.

• Test the model:

python test.py --dataroot ./path/to/testset --name moire_cyclegan --model cycle_gan

• The test results will be saved to a html file here: ./results/moire_cyclegan/latest_test/index.html.

Citation

If you use this code for your research, please cite our papers.

@article{YUE2021,
title = {Unsupervised Moiré Pattern Removal for Recaptured Screen Images},
journal = {Neurocomputing},
year = {2021},
author = {Huanjing Yue and Yijia Cheng and Fanglong Liu and Jingyu Yang}
}

Acknowledgments

Our code is inspired by pytorch-CycleGAN-and-pix2pix.