Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Hu, Shiyu, Li, Xuchen, Li, Xuzhao, Zhang, Jing, Wang, Yipei, Zhao, Xin, Cheong, Kang Hao
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2410.15270
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866908369588060160
author Hu, Shiyu
Li, Xuchen
Li, Xuzhao
Zhang, Jing
Wang, Yipei
Zhao, Xin
Cheong, Kang Hao
author_facet Hu, Shiyu
Li, Xuchen
Li, Xuzhao
Zhang, Jing
Wang, Yipei
Zhao, Xin
Cheong, Kang Hao
contents Despite rapid progress in large vision-language models (LVLMs), existing video caption benchmarks remain limited in evaluating their alignment with human understanding. Most rely on a single annotation per video and lexical similarity-based metrics, failing to capture the variability in human perception and the cognitive importance of events. These limitations hinder accurate diagnosis of model capabilities in producing coherent, complete, and human-aligned descriptions. To address this, we introduce FIOVA (Five-In-One Video Annotations), a human-centric benchmark tailored for evaluation. It comprises 3,002 real-world videos (about 33.6s each), each annotated independently by five annotators. This design enables modeling of semantic diversity and inter-subjective agreement, offering a richer foundation for measuring human-machine alignment. We further propose FIOVA-DQ, an event-level evaluation metric that incorporates cognitive weights derived from annotator consensus, providing fine-grained assessment of event relevance and semantic coverage. Leveraging FIOVA, we conduct a comprehensive evaluation of nine representative LVLMs and introduce a complexity-aware analysis framework based on inter-annotator variation (CV). This reveals consistency gaps across difficulty levels and identifies structural issues such as event under-description and template convergence. Our results highlight FIOVA's diagnostic value for understanding LVLM behavior under varying complexity, setting a new standard for cognitively aligned evaluation in long-video captioning. The benchmark, annotations, metric, and model outputs are publicly released to support future evaluation-driven research in video understanding. More detailed information can be found at https://huuuuusy.github.io/fiova/.
format Preprint
id arxiv_https___arxiv_org_abs_2410_15270
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle FIOVA: A Multi-Annotator Benchmark for Human-Aligned Video Captioning
Hu, Shiyu
Li, Xuchen
Li, Xuzhao
Zhang, Jing
Wang, Yipei
Zhao, Xin
Cheong, Kang Hao
Computer Vision and Pattern Recognition
Despite rapid progress in large vision-language models (LVLMs), existing video caption benchmarks remain limited in evaluating their alignment with human understanding. Most rely on a single annotation per video and lexical similarity-based metrics, failing to capture the variability in human perception and the cognitive importance of events. These limitations hinder accurate diagnosis of model capabilities in producing coherent, complete, and human-aligned descriptions. To address this, we introduce FIOVA (Five-In-One Video Annotations), a human-centric benchmark tailored for evaluation. It comprises 3,002 real-world videos (about 33.6s each), each annotated independently by five annotators. This design enables modeling of semantic diversity and inter-subjective agreement, offering a richer foundation for measuring human-machine alignment. We further propose FIOVA-DQ, an event-level evaluation metric that incorporates cognitive weights derived from annotator consensus, providing fine-grained assessment of event relevance and semantic coverage. Leveraging FIOVA, we conduct a comprehensive evaluation of nine representative LVLMs and introduce a complexity-aware analysis framework based on inter-annotator variation (CV). This reveals consistency gaps across difficulty levels and identifies structural issues such as event under-description and template convergence. Our results highlight FIOVA's diagnostic value for understanding LVLM behavior under varying complexity, setting a new standard for cognitively aligned evaluation in long-video captioning. The benchmark, annotations, metric, and model outputs are publicly released to support future evaluation-driven research in video understanding. More detailed information can be found at https://huuuuusy.github.io/fiova/.
title FIOVA: A Multi-Annotator Benchmark for Human-Aligned Video Captioning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2410.15270