Seeing What Matters: Generalizable AI-generated Video Detection with Forensic-Oriented Augmentation

TL;DR

This paper introduces a forensic-oriented data augmentation method and a training paradigm that significantly enhance the generalization of AI-generated video detectors across diverse generative models, focusing on intrinsic artifacts rather than semantic flaws.

Contribution

The work proposes a novel wavelet-based augmentation strategy and a training approach that improve detector robustness without relying on multiple datasets or complex algorithms.

Findings

Improved detection accuracy across various generative models.

Effective even on recent models like NOVA and FLUX.

Enhanced generalization without large multi-generator datasets.

Abstract

Synthetic video generation is progressing very rapidly. The latest models can produce very realistic high-resolution videos that are virtually indistinguishable from real ones. Although several video forensic detectors have been recently proposed, they often exhibit poor generalization, which limits their applicability in a real-world scenario. Our key insight to overcome this issue is to guide the detector towards *seeing what really matters*. In fact, a well-designed forensic classifier should focus on identifying intrinsic low-level artifacts introduced by a generative architecture rather than relying on high-level semantic flaws that characterize a specific model. In this work, first, we study different generative architectures, searching and identifying discriminative features that are unbiased, robust to impairments, and shared across models. Then, we introduce a novel forensic-oriented data augmentation strategy based on the wavelet decomposition and replace specific frequency-related bands to drive the model to exploit more relevant forensic cues. Our novel training paradigm improves the generalizability of AI-generated video detectors, without the need for complex algorithms and large datasets that include multiple synthetic generators. To evaluate our approach, we train the detector using data from a single generative model and test it against videos produced by a wide range of other models. Despite its simplicity, our method achieves a significant accuracy improvement over state-of-the-art detectors and obtains excellent results even on very recent generative models, such as NOVA and FLUX.