Neural Style Transfer

In learning and experimenting with Convalutional Neural Networks for image recognition, I came across the idea of developing a style transfer. This idea was presented by Fran├žois Chollet in his book Deep Learning with Python . Chollet himself is the author of the Keras package, and a data scientist at Google. To apply the style transfer, small Kernels from each convolution are used to develop a model of the style and localized textures of one image, in this case Picasso's self portrait. A seperate model is then developed for the overall structure of the target image we wish to apply the style to. I then applied the learned style representation onto this to create a stylized transfer!



+







=