Research map / Video / Video-language understanding
AVFakeBench: A Comprehensive Audio-Video Forgery Detection Benchmark for AV-LMMs
Shuhan Xia, Peipei Li, Xuannan Liu, Dongsen Zhang, Xinyu Guo, Zekun Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| evaluate on general-subject instead of human-subject clips | LipFD · AVFakeBench Binary Judgment | F1: 45.0 -> 18.9 | Hurt |
| evaluate on general-subject instead of human-subject clips | AVH-Align · AVFakeBench Binary Judgment | F1: 51.7 -> 34.3 | Hurt |
| evaluate on general-subject instead of human-subject clips | Gemini-2.5-pro · AVFakeBench Binary Judgment | F1: 63.3 -> 54.3 | Hurt |
| use Gemini-2.5-pro AV-LMM instead of expert detector AVH-Align | Gemini-2.5-pro · AVFakeBench Binary Judgment (overall) | macro-F1: 42.8 -> 59.9 | Improved |
| switch task from binary judgment to forgery type classification | Gemini-2.5-pro · AVFakeBench | macro-F1: 59.9 -> 19.2 | Hurt |
| edited forgeries instead of real samples in classification | Gemini-2.5-pro · AVFakeBench Forgery Classification | F1: 42.8 -> 7.5 | Hurt |
| edited forgeries instead of synthesized samples in classification | Gemini-2.5-pro · AVFakeBench Forgery Classification | F1: 13.9 -> 7.5 | Hurt |
| video-only forgery instead of audio-only forgery | GPT-4o · AVFakeBench Forgery Classification | F1: 8.0 -> 21.7 | Improved |
| video-only forgery instead of audio-only forgery | OneLLM · AVFakeBench Forgery Classification | F1: 0.0 -> 18.8 | Improved |
| video-only forgery instead of audio-only forgery | Video-LLaMA2 · AVFakeBench Forgery Classification | F1: 0.0 -> 14.7 | Improved |
| audio-only forgery instead of video-only forgery | Gemini-2.5-pro · AVFakeBench Forgery Classification | F1: 23.0 -> 28.4 | Improved |
| proprietary GPT-4o instead of best open-source video-SALMONN on detail selection | GPT-4o · AVFakeBench Forgery Detail Selection | macro-F1: 17.8 -> 27.5 | Improved |
Closest papers by meaning
- A Hitchhiker's Guide to Fine-Grained Face Forgery Detection Using Common Sense Reasoning (NeurIPS 2024 · similarity 0.89)
- Forensics-Bench: A Comprehensive Forgery Detection Benchmark Suite for Large Vision Language Models (CVPR 2025 · similarity 0.89)
- Unlocking the Capabilities of Large Vision-Language Models for Generalizable and Explainable Deepfake Detection (ICML 2025 · similarity 0.88)
- Exposing the Deception: Uncovering More Forgery Clues for Deepfake Detection (AAAI 2024 · similarity 0.88)
- TriDF: Evaluating Perception, Detection, and Hallucination for Interpretable DeepFake Detection (CVPR 2026 · similarity 0.88)
- Rethinking Vision-Language Model in Face Forensics: Multi-Modal Interpretable Forged Face Detector (CVPR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
