Exposing AI-generated Videos: A Benchmark Dataset and a Local-and-Global Temporal Defect Based Detection Method

TL;DR

This paper introduces a new benchmark dataset of AI-generated videos and proposes a detection method based on analyzing local and global temporal defects, aiming to improve the identification of fake videos.

Contribution

The paper constructs a comprehensive dataset of AI-generated videos using diffusion algorithms and develops a novel detection framework analyzing temporal defects for better fake video detection.

Findings

The dataset enables standardized evaluation of detection methods.

The proposed detection framework shows robustness against degraded videos.

Baseline results highlight challenges in generalization and robustness.

Abstract

The generative model has made significant advancements in the creation of realistic videos, which causes security issues. However, this emerging risk has not been adequately addressed due to the absence of a benchmark dataset for AI-generated videos. In this paper, we first construct a video dataset using advanced diffusion-based video generation algorithms with various semantic contents. Besides, typical video lossy operations over network transmission are adopted to generate degraded samples. Then, by analyzing local and global temporal defects of current AI-generated videos, a novel detection framework by adaptively learning local motion information and global appearance variation is constructed to expose fake videos. Finally, experiments are conducted to evaluate the generalization and robustness of different spatial and temporal domain detection methods, where the results can serve as the baseline and demonstrate the research challenge for future studies.