A while ago we saw this post about a new site by Google called Seedbank. It's goal is to collect machine learning notebooks and examples so that people can get to know machine learning and its capabilities by trying it out. It even offers some free GPU back ends to make the examples execute in reasonable time.
One of the seeds is a notebook on image style transfer based on the famous paper A Neural Algorithm of Artistic Style by Leon A. Gatys et al. Style Transfer got a lot of media attention when it was published because it is a good example of something visually appealing created by machine learning.
We met a lot of family over the holidays and we thought it would be fun to show them some cool things that can be done with machine learning. So we took the code of the notebook and turned it into a script. We than created a container service that allowed anyone within our local network to have their pictures be style transferred.
The interface was copied from one of our older projects. The code is on our github. Here are some style transfer results of our Christmas tree.
Keine Kommentare:
Kommentar veröffentlichen