Would you like to walk inside your favourite picture? Maybe one of Gauguin’s impressionist landscapes, or Dali’s surrealist dreams, or maybe one of David Hockney’s pop art scenes. If you can’t yet, at least your digital memories can now be painted with your preferred artistic style.
In a paper(1) published last year, Leon A. Gatys, Alexander S. Ecker, Matthias Bethge describe an algorithm that uses a deep neural network representation to separate and recombine content and style of arbitrary images, as you can see here:
Now in “Artistic style transfer for videos”(2), Manuel Ruder, Alexey Dosovitskiy, and Thomas Brox extend the approach to transfer the style from one image (e.g, your favourite painting) to a whole video sequence. The result is quite impressive!
It is not too difficult to imagine what’s coming next. Here is the code(3). Let your imagination run wild!
____________________
(1) Gatys, Leon A, Alexander S Ecker, and Matthias Bethge. ‘A Neural Algorithm of Artistic Style’. arXiv Preprint arXiv:1508.06576, 2015.
(2) Ruder, Manuel, Alexey Dosovitskiy, and Thomas Brox. ‘Artistic Style Transfer for Videos’. arXiv Preprint arXiv:1604.08610, 2016.
(3) https://github.com/jcjohnson/neural-style ; https://github.com/manuelruder/artistic-videos
[…] Can machines learn to scare us? Three scientists from the MIT Media Lab—Pinar Yanardag, Manuel Cebrian, Iyad Rahwan—are trying to teach them, using state-of-the-art-now deep learning algorithms: […]