This project aims at extracting the style from a painting and apply it to pictures. The extracted style will act as a filter, like the one you can commonly use on social media.
- Preprocess paintings images
- Implement the VGG neural network
- Apply the output style to an existing picture
The project is based on a study called A Neural Algorithm of Artistic Style, led by Leon Gatys, Alexandre Ecker and Matthias Bethge, researchers from the University of Tübingen in Germany.
Link : https://arxiv.org/pdf/1508.06576.pdf
The dataset used from this project was taken from kaggle.com : it contains multiple impressionists paintings scrapped on WikiArt.
Link : https://www.kaggle.com/delayedkarma/impressionist-classifier-data
The dataset is composed of multiple pictures of 10 famous painters paintings. They are in JPG format. The first step is to visualize them. For this, one should convert the JPG files into numpy arrays, data structure that Python can easily handle.
The first idea is to implement a CNN model to extract the style of a painting. The model will be trained on 10 famous painters artwork in order to extract the 10 corresponding styles. The second part will consist of applying this style to a picture of a day-to-day moment.