Image Style Transfer and Content-Style Disentanglement

TL;DR

This paper introduces a method for learning disentangled content and style representations in images, enabling style extrapolation and interpolation through supervised learning with triplet and cycle-consistency losses.

Contribution

It presents a novel approach combining triplet and cycle-consistency losses to achieve effective content-style disentanglement in images.

Findings

Successful style extrapolation and interpolation demonstrated

Disentangled representations improve image reconstruction fidelity

Supervised data augmentation enhances separation of content and style

Abstract

We propose a way of learning disentangled content-style representation of image, allowing us to extrapolate images to any style as well as interpolate between any pair of styles. By augmenting data set in a supervised setting and imposing triplet loss, we ensure the separation of information encoded by content and style representation. We also make use of cycle-consistency loss to guarantee that images could be reconstructed faithfully by their representation.