Process Over Outcome: Cultivating Forensic Reasoning for Generalizable Multimodal Manipulation Detection

Recent advances in generative AI have significantly enhanced the realism of multimodal media manipulation, thereby posing substantial challenges to manipulation detection. Existing manipulation detection and grounding approaches predominantly focus on manipulation type classification under result-oriented supervision, which not only lacks interpretability but also tends to overfit superficial artifacts. In this paper, we argue that generalizable detection requires incorporating explicit forensic reasoning, rather than merely classifying a limited set of manipulation types, which fails to generalize to unseen manipulation patterns. To this end, we propose REFORM, a reasoning-driven framework that shifts learning from outcome fitting to process modeling. REFORM adopts a three-stage curriculum that first induces forensic rationales, then aligns reasoning with final judgments, and finally refines logical consistency via reinforcement learning. To support this paradigm, we introduce ROM, a large-scale dataset with rich reasoning annotations. Extensive experiments show that REFORM establishes new state-of-the-art performance with superior generalization, achieving 81.52% ACC on ROM, 76.65% ACC on DGM4, and 74.9 F1 on MMFakeBench.

Key Contributions

• REFORM: a three-stage curriculum framework that models forensic reasoning as a process (rationale induction → judgment alignment → RL-based logical consistency refinement) rather than outcome classification
• ROM: a large-scale multimodal manipulation dataset with rich forensic reasoning annotations to support process-oriented supervision
• Demonstrated superior generalization over existing manipulation detection methods, achieving new SOTA on ROM (81.52% ACC), DGM4 (76.65% ACC), and MMFakeBench (74.9 F1)

🛡️ Threat Analysis

Details

Similar Papers