This paper investigates how to effectively utilize Vision-Language Models (VLMs) for image forgery detection and localization (IFDL). It introduces IFDL-VLM, a novel two-stage pipeline that decouples the core IFDL task from VLM-based explanation generation, leveraging localization masks to significantly enhance VLM interpretability and achieving state-of-the-art results across multiple benchmarks.

This paper examines the frequency bias in DNN-based forgery detectors and proposes a frequency alignment method to improve detection reliability.

The paper introduces SDiFL, a novel framework leveraging Stable Diffusion's generative and perceptual capabilities for efficient and accurate image forgery localization. It integrates high-frequency residual signals as an explicit modality within the SD3 latent space, achieving state-of-the-art performance and strong generalization across various forgery types and real-world scenarios.