VRBench: A Benchmark for Multi-Step Reasoning in Long Narrative Videos

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yu, Jiashuo, Wu, Yue, Chu, Meng, Ren, Zhifei, Huang, Zizheng, Chu, Pei, Zhang, Ruijie, He, Yinan, Li, Qirui, Li, Songze, Li, Zhenxiang, Tu, Zhongying, He, Conghui, Qiao, Yu, Wang, Yali, Wang, Yi, Wang, Limin
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909720273485824
author Yu, Jiashuo
Wu, Yue
Chu, Meng
Ren, Zhifei
Huang, Zizheng
Chu, Pei
Zhang, Ruijie
He, Yinan
Li, Qirui
Li, Songze
Li, Zhenxiang
Tu, Zhongying
He, Conghui
Qiao, Yu
Wang, Yali
Wang, Yi
Wang, Limin
author_facet Yu, Jiashuo
Wu, Yue
Chu, Meng
Ren, Zhifei
Huang, Zizheng
Chu, Pei
Zhang, Ruijie
He, Yinan
Li, Qirui
Li, Songze
Li, Zhenxiang
Tu, Zhongying
He, Conghui
Qiao, Yu
Wang, Yali
Wang, Yi
Wang, Limin
contents We present VRBench, the first long narrative video benchmark crafted for evaluating large models' multi-step reasoning capabilities, addressing limitations in existing evaluations that overlook temporal reasoning and procedural validity. It comprises 960 long videos (with an average duration of 1.6 hours), along with 8,243 human-labeled multi-step question-answering pairs and 25,106 reasoning steps with timestamps. These videos are curated via a multi-stage filtering process including expert inter-rater reviewing to prioritize plot coherence. We develop a human-AI collaborative framework that generates coherent reasoning chains, each requiring multiple temporally grounded steps, spanning seven types (e.g., event attribution, implicit inference). VRBench designs a multi-phase evaluation pipeline that assesses models at both the outcome and process levels. Apart from the MCQs for the final results, we propose a progress-level LLM-guided scoring metric to evaluate the quality of the reasoning chain from multiple dimensions comprehensively. Through extensive evaluations of 12 LLMs and 19 VLMs on VRBench, we undertake a thorough analysis and provide valuable insights that advance the field of multi-step reasoning.
format Preprint
id arxiv_https___arxiv_org_abs_2506_10857
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VRBench: A Benchmark for Multi-Step Reasoning in Long Narrative Videos
Yu, Jiashuo
Wu, Yue
Chu, Meng
Ren, Zhifei
Huang, Zizheng
Chu, Pei
Zhang, Ruijie
He, Yinan
Li, Qirui
Li, Songze
Li, Zhenxiang
Tu, Zhongying
He, Conghui
Qiao, Yu
Wang, Yali
Wang, Yi
Wang, Limin
Computer Vision and Pattern Recognition
Artificial Intelligence
Multimedia
We present VRBench, the first long narrative video benchmark crafted for evaluating large models' multi-step reasoning capabilities, addressing limitations in existing evaluations that overlook temporal reasoning and procedural validity. It comprises 960 long videos (with an average duration of 1.6 hours), along with 8,243 human-labeled multi-step question-answering pairs and 25,106 reasoning steps with timestamps. These videos are curated via a multi-stage filtering process including expert inter-rater reviewing to prioritize plot coherence. We develop a human-AI collaborative framework that generates coherent reasoning chains, each requiring multiple temporally grounded steps, spanning seven types (e.g., event attribution, implicit inference). VRBench designs a multi-phase evaluation pipeline that assesses models at both the outcome and process levels. Apart from the MCQs for the final results, we propose a progress-level LLM-guided scoring metric to evaluate the quality of the reasoning chain from multiple dimensions comprehensively. Through extensive evaluations of 12 LLMs and 19 VLMs on VRBench, we undertake a thorough analysis and provide valuable insights that advance the field of multi-step reasoning.
title VRBench: A Benchmark for Multi-Step Reasoning in Long Narrative Videos
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Multimedia
url https://arxiv.org/abs/2506.10857