Forensic-Aware Continual Adaptation for Image Forgery Localization

2026-09-29

AI Summary

This study addresses the limitations of existing image forgery localization methods, which struggle to adapt to emerging manipulation techniques and lack the continual learning capability required to balance old knowledge retention with new domain acquisition over continuous data streams. To this end, this work proposes the first continual learning framework tailored for image forgery localization by establishing the IFL benchmark. It designs spatial mixture-of-forensics-experts and evidence-guided prompting mechanisms to dynamically mine forensic traces, and introduces Fisher-weighted LoRA gradient surgery to mitigate catastrophic forgetting. Experimental results demonstrate that the proposed method achieves state-of-the-art performance in both pixel-level localization and image-level detection across multiple scenarios, enabling effective cross-domain adaptive updates.

Abstract

The rapid evolution of image manipulation techniques has raised growing public security concerns. Existing Image Forgery Localization (IFL) methods can accurately localize manipulated regions but are often unable to adapt to newly emerging forgeries. In real-world forensic scenarios, data typically arrive sequentially, yet continual model adaptation remains largely unexplored in IFL. To bridge this gap, we introduce the first continual learning framework for IFL and establish a comprehensive benchmark under two realistic data-evolution protocols: cross-dataset and cross-content continual learning. Evaluations of representative state-of-the-art IFL and continual learning methods reveal substantial performance degradation, highlighting two key challenges: (1) adaptively capturing intrinsic forensic traces from incoming data across unseen domains, and (2) preserving previously acquired forensic knowledge during sequential adaptation. To address these challenges, we propose a forensic-aware continual adaptation framework. First, a forensic trace mining module employs Spatial Mixture-of-Forensic-Experts (SMoFE) to dynamically route complementary forensic cues across spatial locations, together with Forensic Evidence-Guided Dense Prompting (FEGDP) to transform low-level forensic traces into structured localization evidence for SAM. Second, Fisher-weighted LoRA Gradient (FLAG) surgery identifies old-task-sensitive adaptation directions and suppresses conflicting updates, mitigating catastrophic forgetting while preserving plasticity for emerging forgery domains. Extensive experiments demonstrate state-of-the-art performance in both pixel-level forgery localization and image-level forgery detection across diverse continual learning scenarios.

Problem

• Image Forgery Localization
• Continual Learning
• Catastrophic Forgetting
• Forensic Traces
• Domain Adaptation

Innovation

• Continual Learning
• Image Forgery Localization
• Mixture-of-Forensic-Experts
• Dense Prompting
• LoRA Gradient Surgery