Convnet Style-Transfer

-> This document is based on the famous research paper Gatys et al., 2015

Style transfer is the technique of recomposing images in the style of other images. These were mostly created using paper by Gatys, Ecker, and Bethge demonstrating a method for restyling images using convolutional neural networks.

Requirements

You will need the following to run the above:

• Python3.5
• Tensorflow
• Keras
• Numpy

How to run the code

For Ubuntu users:

$ python3 vgg16.py <content_image_path> <style_image_path>

Results

Result based on 10 iterations:

Observations

The following variations can be seen by changing some of the factors:

• content-style weight ratio see here
• extracting content features from different convnet layers see here

Attributions/Thanks

• Some code was borrowed from Justin Johnson's Fast Neural Style