Back to publications

VQualA 2025 Challenge on GenAI-Bench AIGC Video Quality Assessment: Methods and Results

Ying Cheng, Huasheng Wang, Pengxiang Xiao, Yukang Ding, Enpeng Liu, Chris Wei Zhou, Baojun Li, Jiamian Huang, Jiarui Wang, Xiaorong Zhu, Juntong Wang, Huiyu Duan, Xiongkuo Min, Qiang Hu, Chunlei Cai, Guangtao Zhai, Baihong Qian, Haotian Fan, Wenjie Liao, Yunqiu Wang, Tao Li, Junuhi Cui, Zhichao Zhang, Xinyue Li, Yunhao Li, Xiaohong Liu, Weixia Zhang, Bingkun Zheng, Wei Sun, Zhihua Wang, Longwei Li, Jinyu Zhao, Xincheng Lv, Yang Cai, Fangfang Lu, Ritik Bompilwar, Saurabh Koshatwar, Weifeng Cai, Guangqian Kong, Junfeng Yang, Jing Fu, Wei Zhang, Wenzhi Cao, Limei Liu, Qin Li, Wanli Ma, Yixiao Li, Xiaoshuai Hao

ICCV 2025 · Visual Quality Assessment (VQualA) Workshop
Paper

Abstract

This paper presents a review of the VQualA 2025 Challenge on GenAI-Bench, which focuses on the critical task of AI-Generated Content (AIGC) video quality assessment. The challenge was structured into two distinct tracks to foster comprehensive evaluation methodologies: (i) Track I: Overall Quality Assessment, where participants were required to develop models that predict a single, holistic quality score reflecting overall human perception; and (ii) Track II: Multi Dimensional Quality Assessment, which challenged teams to predict separate scores across four key dimensions: Aesthetic Quality, Image Quality, Temporal Quality, and Text-Video Alignment. Both tracks utilized the newly introduced TaobaoVD-GC dataset, comprising 5,000 videos generated by five advanced text-to-video (T2V) models. The primary objective of the challenge was to establish a standardized benchmark for AIGC video quality and to drive research beyond traditional metrics, addressing unique generative artifacts such as temporal inconsistency and semantic drift. It drew considerable attention, with an average of 85 teams registered for each track, and received 15 valid final submissions with detailed methodology reports.

VQualA 2025 Challenge on GenAI-Bench AIGC Video Quality Assessment

Citation

@InProceedings{Cheng_2025_ICCV,
  author    = {Cheng, Ying and Wang, Huasheng and Xiao, Pengxiang and Ding, Yukang and Liu, Enpeng and Zhou, Chris Wei and Li, Baojun and Huang, Jiamian and Wang, Jiarui and Zhu, Xiaorong and Wang, Juntong and Duan, Huiyu and Min, Xiongkuo and Hu, Qiang and Cai, Chunlei and Zhai, Guangtao and Qian, Baihong and Fan, Haotian and Liao, Wenjie and Wang, Yunqiu and Li, Tao and Cui, Junuhi and Zhang, Zhichao and Li, Xinyue and Li, Yunhao and Liu, Xiaohong and Zhang, Weixia and Zheng, Bingkun and Sun, Wei and Wang, Zhihua and Li, Longwei and Zhao, Jinyu and Lv, Xincheng and Cai, Yang and Lu, Fangfang and Bompilwar, Ritik and Koshatwar, Saurabh and Cai, Weifeng and Kong, Guangqian and Yang, Junfeng and Fu, Jing and Zhang, Wei and Cao, Wenzhi and Liu, Limei and Li, Qin and Ma, Wanli and Li, Yixiao and Hao, Xiaoshuai},
  title     = {VQualA 2025 Challenge on GenAI-Bench AIGC Video Quality Assessment: Methods and Results},
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops},
  month     = {October},
  year      = {2025},
  pages     = {3517-3526}
}