Intelligent Mobile Projects with TensorFlow

Neural Style Transfer – a quick overview

The original idea and algorithm of using a deep neural network to merge the content of an image with the style of another was published in a paper titled A Neural Algorithm of Artistic Style (https://arxiv.org/abs/1508.06576) in the summer of 2015. It was based on a pre-trained deep CNN model called VGG-19 (https://arxiv.org/pdf/1409.1556.pdf), the winner of the 2014 ImageNet image recognition challenge, which has 16 convolutional layers, or feature maps, representing different levels of the image content. In this original method, the final transferred image is first initialized as a white noise image merged with the content image. The content loss function is defined as the squared error loss of a ...