Junshk / CinCGAN-pytorch

Unofficial Implementation of "Unsupervised Image Super-Resolution using Cycle-in-Cycle Generative Adversarial Networks" in CVPR 2018.

CinCGAN-pytorch

This repository is a PyTorch version of the paper "Unsupervised Image Super-Resolution using Cycle-in-Cycle Generative Adversarial Networks" from CVPRW 2018.

For super-resolution setting I refer to EDSR (PyTorch) (you can download pretrained EDSR from here)

This version is not good looking yet. It will be updated later..

---

Train

• Dataset

  Download DIV2K dataset (NTIRE2018) from here. 800 training (800) & 100 validation images (801900)

• Pretrained EDSR network.

  Download pretrained EDSR from here

• Execution

  After move to 'code' folder, type the following command.

   python main.py --dir_data 'data_path'

  data_path directory should contains 'DIV2K' dataset folder.

Test

• Execution

  After move to 'code' folder, type the following command.

   python main.py --n_val 100