TensorFlow Tutorial - Style Transfer Algorithm

 

Notes by Magnus Erik Hvass Pedersen: https://github.com/Hvass-Labs/TensorFlow-Tutorials/blob/master/15_Style_...

We saw in the previous Tutorial #14 how to maximize the feature activations inside a neural network so as to amplify patterns in the input image. This was called DeepDreaming.

This tutorial uses a similar idea but takes two images as input: A content-image and a style-image. We then wish to create a mixed-image which has the contours of the content-image and the colours and texture of the style-image.

This builds on the previous tutorials. You should be familiar with neural networks in general (e.g. Tutorial #01 and #02) and it may also be helpful to be familiar with Tutorial #14 on DeepDream.

Resource Type: