Code-in-the-Loop Forensics: Agentic Tool Use for Image Forgery Detection

TL;DR

The paper introduces ForenAgent, an interactive framework enabling large language models to autonomously generate and refine tools for more flexible and interpretable image forgery detection, bridging high-level semantics and low-level artifacts.

Contribution

It presents a novel multi-round interactive IFD framework with a training pipeline and a new dataset, enhancing tool interaction and reasoning in forgery detection.

Findings

ForenAgent demonstrates emergent tool-use competence in challenging IFD tasks.

The framework achieves more flexible and interpretable forgery analysis.

Experiments validate the effectiveness of the dynamic reasoning loop.

Abstract

Existing image forgery detection (IFD) methods either exploit low-level, semantics-agnostic artifacts or rely on multimodal large language models (MLLMs) with high-level semantic knowledge. Although naturally complementary, these two information streams are highly heterogeneous in both paradigm and reasoning, making it difficult for existing methods to unify them or effectively model their cross-level interactions. To address this gap, we propose ForenAgent, a multi-round interactive IFD framework that enables MLLMs to autonomously generate, execute, and iteratively refine Python-based low-level tools around the detection objective, thereby achieving more flexible and interpretable forgery analysis. ForenAgent follows a two-stage training pipeline combining Cold Start and Reinforcement Fine-Tuning to enhance its tool interaction capability and reasoning adaptability progressively. Inspired by human reasoning, we design a dynamic reasoning loop comprising global perception, local focusing, iterative probing, and holistic adjudication, and instantiate it as both a data-sampling strategy and a task-aligned process reward. For systematic training and evaluation, we construct FABench, a heterogeneous, high-quality agent-forensics dataset comprising 100k images and approximately 200k agent-interaction question-answer pairs. Experiments show that ForenAgent exhibits emergent tool-use competence and reflective reasoning on challenging IFD tasks when assisted by low-level tools, charting a promising route toward general-purpose IFD. The code will be released after the review process is completed.