Final Degree Project to apply impressionistic style on pictures using Neural Networks.
This repo contains my Final Degree Project of Computer Engineering, Generation of impressionistic pictures using Neural Networks, done on October 2020. This Final Degree Project aims to bring the reader closer to Artificial Intelligence through an artistic perspective.
For this objective, the paintings of three great impressionist and post-impressionist artists have been taken, such as Claude Monet, Vincent Van Gogh and Paul Cézanne; aiming to develop a system based on Neural Networks that transforms photographs into pictures that the reader himself can use on his computer.
- Implementation of CycleGAN neural network on Keras and Tensorflow 2.
- Config files used on my thesis.
- Weigths of trained neural networks.
- For cezzane2photo and monet2photo, saved model in .h5 format with model optimizer state to resume training, architecture and config. For vangogh2photo was not possible due to size limits.
- Whole ocumentation presented: book (LaTeX and pdf) and presentation (pptx and pdf) (only on Spanish). All documents were written only on Spanish.
Upscale system can be found here