InfoStyler: Disentanglement Information Bottleneck for Artistic Style Transfer

TL;DR

InfoStyler introduces a novel disentanglement information bottleneck approach for artistic style transfer, effectively balancing content preservation and style transfer by disentangling content and style features through information compression.

Contribution

The paper proposes a new information disentanglement method called InfoStyler that captures minimal sufficient content and style information, improving style transfer quality.

Findings

Produces high-quality stylized images with balanced content and style.

Effectively disentangles content and style features using information compression.

Outperforms prior methods in style transfer benchmarks.

Abstract

Artistic style transfer aims to transfer the style of an artwork to a photograph while maintaining its original overall content. Many prior works focus on designing various transfer modules to transfer the style statistics to the content image. Although effective, ignoring the clear disentanglement of the content features and the style features from the first beginning, they have difficulty in balancing between content preservation and style transferring. To tackle this problem, we propose a novel information disentanglement method, named InfoStyler, to capture the minimal sufficient information for both content and style representations from the pre-trained encoding network. InfoStyler formulates the disentanglement representation learning as an information compression problem by eliminating style statistics from the content image and removing the content structure from the style image. Besides, to further facilitate disentanglement learning, a cross-domain Information Bottleneck (IB) learning strategy is proposed by reconstructing the content and style domains. Extensive experiments demonstrate that our InfoStyler can synthesize high-quality stylized images while balancing content structure preservation and style pattern richness.