V-STaR Benchmark

What is V-STaR?

V-STaR is a spatio-temporal reasoning benchmark for Video-LLMs, evaluating Video-LLM’s spatio-temporal reasoning ability in answering questions explicitly in…

Released

2025-03-14

Modality

multimodal

Evaluates

多模态, Multimodal, 多模态模型, VLM, 跨模态推理, Cross-modal Reasoning, 视频理解, Video Understanding, 不支持, Unsupported

Openness

restricted

Importer

OpenCompass Hub

Reported scores

0

Source provenance

• Original evidence https://hub.opencompass.org.cn/dataset-detail/V-STaR

V-STaR paper, code and dataset

• Paper https://arxiv.org/abs/2503.11495
• Code https://github.com/V-STaR-Bench/V-STaR
• Dataset https://huggingface.co/datasets/V-STaR-Bench/V-STaR

Which sources cite V-STaR?

• V-STaR

Related 多模态, Multimodal, 多模态模型, VLM, 跨模态推理, Cross-modal Reasoning, 视频理解, Video Understanding, 不支持, Unsupported benchmarks

• 3DCodeBench
• A-Bench
• A-OKVQA
• AbsPyramid
• ActionAtlas
• ActiveFly-Bench
• Ada-LEval
• AECBench