Tutorial on how to run Neural Style Transfer on Google Colab
I am using this implementation https://github.com/jeremycochoy/style... (jupyter notebook + PyTorch) which works flawlessly on Google Colab. The implementation seems to be correct. I went through quite a few implementations of Neural Style. The best one is the one by Justin Johnson on GitHub. This one is very convenient because it's a Jupyter Notebook that can be run on Google Colab using GPU. Takes about 5 minutes for 1000 iterations (512x512 images). In this video, I show you the best parameters and the best layers to choose for the content and style layers. For more painting and 3d stuff (lots of free software for depth maps, etc), check http://3dstereophoto.blogspot.com/.