OmniVideoBench: Audio-Visual Understanding Evaluation for Multimodal LLMs

NJU-LINK's OmniVideoBench is a large-scale benchmark dataset designed to evaluate multimodal large language models on joint audio and visual reasoning tasks. It addresses a gap in existing benchmarks that often focus on a single modality. The dataset was last updated on April 8, 2026.

Use Cases

• Benchmarking model performance on joint audio-visual reasoning tasks based on the described evaluation focus.
• Training multimodal LLMs for video understanding based on the dataset's large-scale, curated nature.
• Studying the interaction between audio and visual modalities in AI systems based on the dataset's stated purpose.

Strengths

• Designed as a large-scale benchmark, suggesting substantial volume.
• Rigorously curated, implying a level of quality control.
• Explicitly targets joint audio and visual reasoning, filling a described gap in evaluation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.
• The description is incomplete, requiring a visit to the external page for full details.

Provenance

Source

NJU-LINK

Freshness

Last updated 2026-04-08 05:19:52.