Advance Fake Video Detection via Vision Transformers
This addresses the urgent need for accurate fake video detection to combat misinformation, but it is incremental as it adapts existing ViT-based image detection to video.
The paper tackles the problem of detecting AI-generated fake videos by extending Vision Transformer (ViT) methods from images to video, proposing a framework that integrates ViT embeddings over time, and reports promising accuracy, generalization, and few-shot learning on a new large dataset from five open-source and proprietary generative techniques.
Recent advancements in AI-based multimedia generation have enabled the creation of hyper-realistic images and videos, raising concerns about their potential use in spreading misinformation. The widespread accessibility of generative techniques, which allow for the production of fake multimedia from prompts or existing media, along with their continuous refinement, underscores the urgent need for highly accurate and generalizable AI-generated media detection methods, underlined also by new regulations like the European Digital AI Act. In this paper, we draw inspiration from Vision Transformer (ViT)-based fake image detection and extend this idea to video. We propose an {original} %innovative framework that effectively integrates ViT embeddings over time to enhance detection performance. Our method shows promising accuracy, generalization, and few-shot learning capabilities across a new, large and diverse dataset of videos generated using five open source generative techniques from the state-of-the-art, as well as a separate dataset containing videos produced by proprietary generative methods.