AVFakeBench: New Benchmark for Audio-Video Forgery Detection in AV-LMMs
Published:Nov 26, 2025 10:33
•1 min read
•ArXiv
Analysis
This ArXiv paper introduces AVFakeBench, a new benchmark designed to evaluate audio-video forgery detection capabilities in Audio-Video Large Language Models (AV-LMMs). The benchmark likely offers a standardized method for assessing and comparing the performance of different AV-LMMs in identifying manipulated content.
Key Takeaways
- •AVFakeBench is a new benchmark for evaluating audio-video forgery detection.
- •It targets AV-LMMs.
- •The benchmark likely standardizes the evaluation process.
Reference
“The paper focuses on creating a benchmark for AV-LMMs.”