DeepObjStyle: Deep Object-based Photo Style Transfer

TL;DR

DeepObjStyle introduces an object-based style transfer method that preserves object semantics and handles mismatched style and content images, improving visual quality without relying on training data.

Contribution

It proposes a novel object-based supervision approach for style transfer that effectively manages semantic mismatches and enhances output quality.

Findings

Outperforms existing methods in preserving object semantics.

Effective in style transfer with mismatched style and content images.

Validated through quantitative metrics and user studies.

Abstract

One of the major challenges of style transfer is the appropriate image features supervision between the output image and the input (style and content) images. An efficient strategy would be to define an object map between the objects of the style and the content images. However, such a mapping is not well established when there are semantic objects of different types and numbers in the style and the content images. It also leads to content mismatch in the style transfer output, which could reduce the visual quality of the results. We propose an object-based style transfer approach, called DeepObjStyle, for the style supervision in the training data-independent framework. DeepObjStyle preserves the semantics of the objects and achieves better style transfer in the challenging scenario when the style and the content images have a mismatch of image features. We also perform style transfer of images containing a word cloud to demonstrate that DeepObjStyle enables an appropriate image features supervision. We validate the results using quantitative comparisons and user studies.