New Omni-Fake dataset benchmarks multimodal deepfake detection on social media

Researchers have introduced Omni-Fake, a new benchmark dataset designed to improve the detection of multimodal deepfakes on social media. The dataset includes over 1 million samples across image, audio, video, and audio-video talking head modalities, along with an out-of-distribution benchmark to test generalization. Omni-Fake also supports a protocol for joint detection, localization, and explanation of deepfakes, and introduces a reinforcement-learning-based detector called Omni-Fake-R1 that integrates cross-modal cues for more accurate and explainable results. AI

IMPACT Enhances the ability to detect sophisticated multimodal deepfakes, crucial for maintaining information integrity on social media platforms.

RANK_REASON The cluster describes a new academic paper introducing a benchmark dataset and a novel detection method for multimodal deepfakes. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• audio-video talking head
• image
• Omni-Fake
• reinforcement learning
• social media
• video
• audio
• deepfake detection

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →