You can deploy this model and start earning money today!

opencompass/
MMBench-Video

visual-question-answering

large vision-language model

Views

No views yet

MMBench-Video: A Long-Form Multi-Shot Benchmark for Holistic Video Understanding

Homepage: https://mmbench-video.github.io/ Repository: https://huggingface.co/datasets/opencompass/MMBench-Video Paper: MMBench-Video: A Long-Form Multi-Shot Benchmark for Holistic Video Understanding.

Introduction

MMBench-Video is a quantitative benchmark designed to rigorously evaluate LVLMs' proficiency in video understanding. MMBench-Video incorporates approximately 600 web videos… See the full description on the dataset page: https://huggingface.co/datasets/opencompass/MMBench-Video.