Explainable Forensics of Manipulated Segments in Untrimmed Long Videos
Researchers have developed a new method for detecting and explaining manipulated segments within long, untrimmed videos, addressing the limitations of existing tools that primarily focus on short clips. This approach, called MSLoc, uses a coarse-to-fine strategy to efficiently scan long videos and then refine the localization and reasoning for AI-generated content. The accompanying TASLE benchmark, featuring over 12,000 videos, provides temporal boundaries, authenticity labels, and rationales for manipulated segments, aiming to improve the analysis of AI-generated misinformation. AI
IMPACT Enhances detection of AI-generated misinformation in long-form video content.