In the light of feature distributions: moment matching for Neural Style Transfer
This work improves style transfer for image processing applications, but it is incremental as it builds on existing distribution-matching concepts.
The paper tackles the problem of Neural Style Transfer by addressing limitations in existing methods that only partially align feature distributions, proposing a novel approach that matches distributions more precisely using Central Moment Discrepancy, resulting in visually better style transfer and improved disentanglement of style from content.
Style transfer aims to render the content of a given image in the graphical/artistic style of another image. The fundamental concept underlying NeuralStyle Transfer (NST) is to interpret style as a distribution in the feature space of a Convolutional Neural Network, such that a desired style can be achieved by matching its feature distribution. We show that most current implementations of that concept have important theoretical and practical limitations, as they only partially align the feature distributions. We propose a novel approach that matches the distributions more precisely, thus reproducing the desired style more faithfully, while still being computationally efficient. Specifically, we adapt the dual form of Central Moment Discrepancy (CMD), as recently proposed for domain adaptation, to minimize the difference between the target style and the feature distribution of the output image. The dual interpretation of this metric explicitly matches all higher-order centralized moments and is therefore a natural extension of existing NST methods that only take into account the first and second moments. Our experiments confirm that the strong theoretical properties also translate to visually better style transfer, and better disentangle style from semantic image content.