MVQA-68K: A Multi-dimensional and Causally-annotated Dataset with Quality Interpretability for Video Assessment

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Pu, Yanyun, Li, Kehan, Huang, Zeyi, Zhong, Zhijie, Yang, Kaixiang
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915494754254848
author Pu, Yanyun
Li, Kehan
Huang, Zeyi
Zhong, Zhijie
Yang, Kaixiang
author_facet Pu, Yanyun
Li, Kehan
Huang, Zeyi
Zhong, Zhijie
Yang, Kaixiang
contents With the rapid advancement of video generation models such as Sora, video quality assessment (VQA) is becoming increasingly crucial for selecting high-quality videos from large-scale datasets used in pre-training. Traditional VQA methods, typically producing single numerical scores, often lack comprehensiveness and interpretability. To address these challenges, we introduce MVQA-68K, a novel multi-dimensional VQA dataset comprising over 68,000 carefully annotated videos, covering seven essential quality dimensions: overall aesthetics, camera movement, dynamic degree, texture detail, composition, visual quality, and factual consistency. Each annotation includes detailed chain-of-thought reasoning to facilitate interpretability and comprehensive understanding. Extensive experiments demonstrate that MVQA-68K significantly enhances the performance of various multimodal large language models (MLLMs) on the VQA task, achieving state-of-the-art results not only on our internal test set (Fig.1) but also on public benchmarks including LSVQ-test, LSVQ-1080p, and LIVE-VQC. Meantime, incorporating explicit reasoning process during VQA training substantially boosts the zero-shot generalization. Code and dataset will be available at github: https://github.com/Controller01-ai/MVQA-68K
format Preprint
id arxiv_https___arxiv_org_abs_2509_11589
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MVQA-68K: A Multi-dimensional and Causally-annotated Dataset with Quality Interpretability for Video Assessment
Pu, Yanyun
Li, Kehan
Huang, Zeyi
Zhong, Zhijie
Yang, Kaixiang
Computer Vision and Pattern Recognition
With the rapid advancement of video generation models such as Sora, video quality assessment (VQA) is becoming increasingly crucial for selecting high-quality videos from large-scale datasets used in pre-training. Traditional VQA methods, typically producing single numerical scores, often lack comprehensiveness and interpretability. To address these challenges, we introduce MVQA-68K, a novel multi-dimensional VQA dataset comprising over 68,000 carefully annotated videos, covering seven essential quality dimensions: overall aesthetics, camera movement, dynamic degree, texture detail, composition, visual quality, and factual consistency. Each annotation includes detailed chain-of-thought reasoning to facilitate interpretability and comprehensive understanding. Extensive experiments demonstrate that MVQA-68K significantly enhances the performance of various multimodal large language models (MLLMs) on the VQA task, achieving state-of-the-art results not only on our internal test set (Fig.1) but also on public benchmarks including LSVQ-test, LSVQ-1080p, and LIVE-VQC. Meantime, incorporating explicit reasoning process during VQA training substantially boosts the zero-shot generalization. Code and dataset will be available at github: https://github.com/Controller01-ai/MVQA-68K
title MVQA-68K: A Multi-dimensional and Causally-annotated Dataset with Quality Interpretability for Video Assessment
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2509.11589