This is my work on Vincent. Vincent is an AI Artist. A Convolutional Neural Network that can extract 'style' and 'content' from artistic images. This immature implementation is based on the pioneer paper "A Neural Algorithm of Artistic Style" (http://arxiv.org/pdf/1508.06576v2.pdf). With only 6 hidden layers, 10 iterations produced this set of output images. My final attempt will be with 31 hidden layers. But to run a CNN with that amount of hidden layers, on a 512*512 size image, it will take a powerful GPU which I do not have at this moment. So I'm asking anybody with a powerful enough GPU to run the code and provide me the outputs. Here's the git repo for Vincent - AI Artist. https://github.com/saikatbsk/Vincent-AI-Artist
Add a comment...