Artistic Style Transfer in Keras

This is a Keras implementation of A Neural Algorithm of Artistic Style by Leon A. Gatys, Alexander S. Ecker and Matthias Bethge.

Neural Styler lets you create artistic images by combining a base picture with the style of another. For example, the images above show multiple iterations of the Chicago skyline combined with Edvard Munch's The Scream.

API

Styling an image is done through generate.py as follows:

python generate.py examples/bases/chicago.jpg examples/styles/umbrella_girl.jpg examples/results/my_result

For a more detailed documentation along with the default parameters click here.

Requirements

keras (and associated dependencies)
- numpy, scipy
- pyyaml
- HDF5 and h5py
pillow
Python 2 and 3

Attribution

This implementation uses some code from Francois Chollet's Neural Style Transfer.
The hierarchy also borrows from Giuseppe's gist which you can view here.
Documentation has also borrowed from Logan Engstrom's Fast Style Transfer