LocateEdit-Bench: A Benchmark for Instruction-Based Editing Localization

TL;DR

LocateEdit-Bench introduces a large-scale dataset for evaluating image forgery localization methods against instruction-based edits, addressing a gap in current AI forgery detection research.

Contribution

The paper presents a new benchmark dataset with diverse editing models and types, along with evaluation protocols, to advance localization methods for instruction-driven image editing.

Findings

Existing methods are ineffective against instruction-based edits.

LocateEdit-Bench provides comprehensive data for benchmarking.

Evaluation protocols enable consistent assessment of localization techniques.

Abstract

Recent advancements in image editing have enabled highly controllable and semantically-aware alteration of visual content, posing unprecedented challenges to manipulation localization. However, existing AI-generated forgery localization methods primarily focus on inpainting-based manipulations, making them ineffective against the latest instruction-based editing paradigms. To bridge this critical gap, we propose LocateEdit-Bench, a large-scale dataset comprising $231$K edited images, designed specifically to benchmark localization methods against instruction-driven image editing. Our dataset incorporates four cutting-edge editing models and covers three common edit types. We conduct a detailed analysis of the dataset and develop two multi-metric evaluation protocols to assess existing localization methods. Our work establishes a foundation to keep pace with the evolving landscape of image editing, thereby facilitating the development of effective methods for future forgery localization. Dataset will be open-sourced upon acceptance.