Skip to content

Dataset · SinerChen

Long video scene understanding benchmark

SinerChen/SceneBench

A collection of videos and questions in English for evaluating scene event understanding by multimodal LLMs.

What's inside

It contains long video clips and related text questions (visual-question-answering) in English. The dataset size is less than 1,000 examples. It is designed to test how well models capture the meaning of entire scenes rather than just recognizing individual frames.

What you can do with it

  • Evaluating multimodal LLMs on long video understanding tasks
  • Comparing video analytics models on their ability to track scene changes
Tasks
Visual question answering
Languages
English
Size
up to 1K rows
Downloads this month
21.2K
Likes
3

How to load it

from datasets import load_dataset

ds = load_dataset("SinerChen/SceneBench")