ManiVid improves detection and explanation of manipulated videos
ManiVid: Unified and Explainable Forensic Analysis of Manipulated Videos
Computer Vision and Pattern Recognition
Summary
Videos can be secretly altered to show things that didn't happen, making it hard to tell what's real. The authors created a big dataset with many examples of these altered videos and a tool called ManiVidLens that can spot manipulated areas, explain what is wrong, and say if a video is real or fake. This tool does better than previous methods at finding the exact parts changed and explaining abnormalities, while still accurately detecting forgeries. This makes it easier to understand and trust the results of video checks.
What this means in practice
- •For media verification teams: Analyze videos to pinpoint manipulated regions and understand anomalies using ManiVidLens to improve trust in news and social media content.
- •For digital forensic investigators: Use ManiVid-38K dataset and ManiVidLens framework to train and evaluate forensic tools that detect and explain video forgeries with high precision.
Authors
Hengrui Kang, Zhonghao Yan, Yuxuan Yang, Ruoyan Jing, Yuncheng Guo, Hao Chen, Kongming Liang, Zhanyu Ma, Conghui He, Weijia Li
Abstract
Rapid advances in AI-generated video (AIGV) have increased the risks posed by deceptive video manipulation. Unlike fully synthetic videos, manipulated videos retain most source content and alter only localized regions, making forensic analysis particularly challenging. Existing video forgery research faces two limitations in both data and methodology: (1) High-quality datasets and benchmarks tailored for manipulated videos remain scarce. (2) Multimodal large language models (MLLMs) extend forgery analysis beyond binary classification but struggle to use low-level forensic cues and provide precise pixel-level grounding. Specifically, we introduce ManiVid, a unified forensic analysis task covering forgery detection, artifact grounding, and anomaly explanation for manipulated videos. We construct ManiVid-38K, the first dataset to combine paired, open-vocabulary localized manipulations of general videos with authenticity labels, forgery masks, and anomaly explanations. It comprises about 19K manually verified real-fake video pairs, mostly at 1080P resolution, generated under 2 paradigms with 15 powerful generation models. We sample 1K pairs for ManiVidBench, balanced across six manipulation types and generation models for fair evaluation. We further propose ManiVidLens, a unified framework for explainable video forgery analysis. Its Forensic Evidence Router supplies shared low-level forensic evidence for multimodal reasoning and video segmentation. Its Prompt Distill Module converts grounding states into semantic and geometric prompts and distills spatial priors for mask decoding and full-video propagation. ManiVidLens achieves relative gains over the strongest comparison methods in artifact grounding (+21.1% mIoU; +21.3% J&F) and anomaly explanation (+131.3% ROUGE-L; +9.9% CSS). Its forgery detection remains comparable to dedicated classifiers (0.914 Acc; 0.913 F1).