Cartoon GAN

Simon Arvidsson and Filip Andersson

This project takes on the problem of transferring the style of cartoon images to real-life photographic images by implementing previous work done by CartoonGAN. Simon and Filip has trained a Generative Adversarial Network (GAN) on over 60 000 images based on the works of Hayao Miyazaki at Studio Ghibli.

A railroad being transformed into a cartoonish style of graphics
Multiple images being transformed form a regular photo to a cartoonish style

gan results

More information about the network architecture and the experiments can be found at their publication. Länk till annan webbplats, öppnas i nytt fönster. 


The source code and the trained models can be found at the Github Repository. Länk till annan webbplats, öppnas i nytt fönster.

Simon and Filip have created a cartoon view of the Jönköping city using their trained model which looks like below.

You can contact to Simon and Filip at;

Simon Arvidsson:

Filip Andersson: