Skip to content

Latest commit

 

History

History
27 lines (17 loc) · 987 Bytes

README.md

File metadata and controls

27 lines (17 loc) · 987 Bytes
Raw

CWGAN

Conditional Wasserstein Generative Adversarial Network for image-to-image translation.

Implementation of the paper: Single Image Haze Removal Using Conditional Wasserstein Generative Adversarial Networks.

Please cite the following work if you use this code:

Single Image Haze Removal Using Conditional Wasserstein Generative Adversarial Networks J.P. Ebenezer, B. Das, S. Mukhopadhyay 2019 27th European Signal Processing Conference (EUSIPCO), 1-5

Arxiv link: https://arxiv.org/pdf/1903.00395.pdf

Training a model

  1. Clone/download the repo
  2. Go to ./scripts/
  3. Change the database location and the other options in train_pix2pix.sh and execute it.

Testing a model

  1. After training the model, go to ./scripts/
  2. Change the database location and the other options in test_pix2pix.sh and execute it.

Acknowledgements

This code is based on https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix (for cGAN) and https://github.com/caogang/wgan-gp (for wGAN).