Skip to content

Fog Simulation using Generative Adversarial Networks (GAN). This code is the implementation of the master thesis Simulating Weather Conditions on Digital Images. It uses a modified CycleGAN model to synthesize fog on clear images.

License

Notifications You must be signed in to change notification settings

josh3255/Foggy-CycleGAN

 
 

Repository files navigation

Foggy-CycleGAN

This project is the implementation for my Computer Science MSc thesis in the University of Debrecen.

Dissertation: [PDF] Simulating Weather Conditions on Digital Images (Debrecen, 2020).

Description

Foggy-CycleGAN is a CycleGAN model trained to synthesize fog on clear images. More details in the dissertation above.

Code

The full source code is available under GPL-3.0 License in my Github repository ghaiszaher/Foggy-CycleGAN

Pre-trained Models

A version of pre-trained models used in the thesis can be found here.

Notebook Open In Colab

A Jupyter Notebook file Foggy_CycleGAN.ipynb is available in the repository.

Results

© Ghais Zaher 2020

About

Fog Simulation using Generative Adversarial Networks (GAN). This code is the implementation of the master thesis Simulating Weather Conditions on Digital Images. It uses a modified CycleGAN model to synthesize fog on clear images.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 77.4%
  • Jupyter Notebook 22.3%
  • Shell 0.3%