0

Consistent Style Transfer

The progressive attentional manifold alignment (PAMA) method improves style transfer by aligning content and style manifolds through repeated attention operations and space-aware interpolations, ensuring consistent semantic region stylization.

Year
2022
Venue
arXiv 2022
Authors
4
Hosting
Abstract onlyARXIV-DEFAULT

Cite

Notes

Only stored in your browser.

Attribution

Abstract & full text
arxiv.org/abs/2201.02233ARXIV-DEFAULT
TL;DR
Semantic Scholar
Attribution policy →

Abstract

Recently, attentional arbitrary style transfer methods have been proposed to achieve fine-grained results, which manipulates the point-wise similarity between content and style features for stylization. However, the attention mechanism based on feature points ignores the feature multi-manifold distribution, where each feature manifold corresponds to a semantic region in the image. Consequently, a uniform content semantic region is rendered by highly different patterns from various style semantic regions, producing inconsistent stylization results with visual artifacts. We proposed the progressive attentional manifold alignment (PAMA) to alleviate this problem, which repeatedly applies attention operations and space-aware interpolations. The attention operation rearranges style features dynamically according to the spatial distribution of content features. This makes the content and style manifolds correspond on the feature map. Then the space-aware interpolation adaptively interpolates between the corresponding content and style manifolds to increase their similarity. By gradually aligning the content manifolds to style manifolds, the proposed PAMA achieves state-of-the-art performance while avoiding the inconsistency of semantic regions. Codes are available at https://github.com/computer-vision2022/PAMA.

Authors

4