Style Spectroscope: Improve Interpretability and Controllability through Fourier Analysis

08/12/2022
by   Zhiyu Jin, et al.
19

Universal style transfer (UST) infuses styles from arbitrary reference images into content images. Existing methods, while enjoying many practical successes, are unable of explaining experimental observations, including different performances of UST algorithms in preserving the spatial structure of content images. In addition, methods are limited to cumbersome global controls on stylization, so that they require additional spatial masks for desired stylization. In this work, we provide a systematic Fourier analysis on a general framework for UST. We present an equivalent form of the framework in the frequency domain. The form implies that existing algorithms treat all frequency components and pixels of feature maps equally, except for the zero-frequency component. We connect Fourier amplitude and phase with Gram matrices and a content reconstruction loss in style transfer, respectively. Based on such equivalence and connections, we can thus interpret different structure preservation behaviors between algorithms with Fourier phase. Given the interpretations we have, we propose two manipulations in practice for structure preservation and desired stylization. Both qualitative and quantitative experiments demonstrate the competitive performance of our method against the state-of-the-art methods. We also conduct experiments to demonstrate (1) the abovementioned equivalence, (2) the interpretability based on Fourier amplitude and phase and (3) the controllability associated with frequency components.

READ FULL TEXT

page 7

page 8

page 9

page 10

research
04/09/2019

Multimodal Style Transfer via Graph Cuts

An assumption widely used in recent neural style transfer methods is tha...
research
07/06/2019

Fast Universal Style Transfer for Artistic and Photorealistic Rendering

Universal style transfer is an image editing task that renders an input ...
research
08/10/2021

Domain-Aware Universal Style Transfer

Style transfer aims to reproduce content images with the styles from ref...
research
03/04/2023

Decompose, Adjust, Compose: Effective Normalization by Playing with Frequency for Domain Generalization

Domain generalization (DG) is a principal task to evaluate the robustnes...
research
03/17/2022

Depth-aware Neural Style Transfer using Instance Normalization

Neural Style Transfer (NST) is concerned with the artistic stylization o...
research
03/31/2018

Quantitative Evaluation of Style Transfer

Style transfer methods produce a transferred image which is a rendering ...
research
03/23/2022

Deep Frequency Filtering for Domain Generalization

Improving the generalization capability of Deep Neural Networks (DNNs) i...

Please sign up or login with your details

Forgot password? Click here to reset