Inharmonious Region Localization by Magnifying Domain Discrepancy

TL;DR

This paper introduces a novel framework that magnifies domain discrepancies in color space to improve localization of inharmonious regions in synthetic images, demonstrating superior performance on relevant datasets.

Contribution

The work proposes a new color space transformation and a domain discrepancy magnification loss for better inharmonious region localization, advancing the state of the art.

Findings

Outperforms existing methods on image harmonization datasets

Effective in magnifying domain discrepancies for localization

Flexible framework with arbitrary localization network

Abstract

Inharmonious region localization aims to localize the region in a synthetic image which is incompatible with surrounding background. The inharmony issue is mainly attributed to the color and illumination inconsistency produced by image editing techniques. In this work, we tend to transform the input image to another color space to magnify the domain discrepancy between inharmonious region and background, so that the model can identify the inharmonious region more easily. To this end, we present a novel framework consisting of a color mapping module and an inharmonious region localization network, in which the former is equipped with a novel domain discrepancy magnification loss and the latter could be an arbitrary localization network. Extensive experiments on image harmonization dataset show the superiority of our designed framework. Our code is available at https://github.com/bcmi/MadisNet-Inharmonious-Region-Localization.