视频动作质量评估  

Video action quality assessment

作　　者：曾令安 郑伟诗[1] ZENG Lingan;ZHENG Weishi(Sun Yat-Sen University,Guangzhou 510275,China)

机构地区：[1]中山大学,广州510275

出　　处：《中国传媒大学学报（自然科学版）》2021年第5期29-35,共7页Journal of Communication University of China：Science and Technology

基　　金：国家自然科学基金(U1911401,U1811461)。

摘　　要：视频动作质量评估旨在通过计算机对视频中人体行为进行分数评估,其可以作为监测和评估人体行为质量的自动化技术,因而在实验操作规范监测、康复医疗和体育训练等领域有着重要的应用价值。本文介绍了视频动作质量评估任务的研究挑战,而后梳理了相关的研究工作,介绍了传统方法的和深度学习方法,并详细分析了这两类方法的优缺点,最后介绍了其在现实生活中的应用价值。Video action quality assessment aims to evaluate human behavior in a video by the computer. Action quality assessment can be used as an automated technology for real-time monitoring and evaluating the quality of human action, and has important application value in the fields of experimental monitoring, rehabilitation medicine, and sports. This survey introduces the challenges of video action quality assessment tasks. Then,this survey systematically reviews the related methods and introduces traditional methods and deep learning methods with analyzing the pros and cons of these methods. Finally, this survey concludes the application in real life.

关 键 词：视频动作质量评估 视频理解 

分 类 号：O422[理学—声学]