VQualA 2025短视频参与度预测挑战赛：方法与结果概述
计算机视觉与模式识别
2025-09-04 v1 多媒体
社会与信息网络
摘要
本文介绍了VQualA 2025短视频参与度预测挑战赛概况，该挑战赛于ICCV 2025上举办。挑战赛聚焦于理解和建模社交媒体平台上用户生成内容(UGC)短视频的流行性。为支持此目标，挑战赛采用新的短视频UGC数据集，包含来自真实用户互动的参与度指标。挑战赛旨在推动捕捉影响用户参与度的复杂因素的鲁棒建模策略。参赛者探索了多种多模态特征，包括视觉内容、音频及创作者提供的元数据。该挑战赛吸引了97名参赛者，收到15份有效测试提交，显著推动了短视频UGC参与度预测领域的进展。
引用
@article{arxiv.2509.02969,
title = {VQualA 2025 Challenge on Engagement Prediction for Short Videos: Methods and Results},
author = {Dasong Li and Sizhuo Ma and Hang Hua and Wenjie Li and Jian Wang and Chris Wei Zhou and Fengbin Guan and Xin Li and Zihao Yu and Yiting Lu and Ru-Ling Liao and Yan Ye and Zhibo Chen and Wei Sun and Linhan Cao and Yuqin Cao and Weixia Zhang and Wen Wen and Kaiwei Zhang and Zijian Chen and Fangfang Lu and Xiongkuo Min and Guangtao Zhai and Erjia Xiao and Lingfeng Zhang and Zhenjie Su and Hao Cheng and Yu Liu and Renjing Xu and Long Chen and Xiaoshuai Hao and Zhenpeng Zeng and Jianqin Wu and Xuxu Wang and Qian Yu and Bo Hu and Weiwei Wang and Pinxin Liu and Yunlong Tang and Luchuan Song and Jinxi He and Jiaru Wu and Hanjia Lyu},
journal= {arXiv preprint arXiv:2509.02969},
year = {2025}
}
备注
ICCV 2025 VQualA workshop EVQA track