Search papers, labs, and topics across Lattice.
This paper introduces PATE-Forensics, a novel approach to explainable deepfake forensics that separates detection and localization from explanation generation, enhancing the interpretability of forensic outputs. By utilizing a DINOv3-based tool that integrates multi-granularity detection with cue-guided localization, the method spatializes evidence into forgery score maps that inform dense mask predictions. The approach achieves a top score of 0.89 on the DDL-X Track 3, significantly surpassing the second-place team by 0.19 points, demonstrating its effectiveness in forensic tasks.
By decoupling detection from explanation, PATE-Forensics achieves unprecedented accuracy in deepfake forensics, setting a new benchmark for explainability in AI.
Existing explainable deepfake forensic methods typically rely on task-adapted MLLM to jointly address detection, localization, and explanation. Inspired by agent-style tool use, we instead introduce a Perception-as-Tool paradigm and instantiate it as PATE-Forensics, which architecturally decouples detection and localization from explanation generation while coupling detection and localization as tightly as possible within a forensic perception tool. The DINOv3-based tool couples a multi-granularity detection module that integrates global, patch-level, and segment-level evidence with a cue-guided localization module by spatializing the patch-level and segment-level evidence into forgery score maps that guide dense mask prediction. The original image and forensic perception outputs produced by the tool form structured forensic context for a general-purpose MLLM, which is guided by prompt constraints to generate explanations without task-specific fine-tuning. On DDL-X Track 3, PATE-Forensics achieves the best official score of 0.89, outperforming the second-ranked team by 0.19 points. Our code is available at https://github.com/yqli00000/PATE-Forensics.
