
软件学报 ISSN 1000-9825, CODEN RUXUEW E-mail: jos@iscas.ac.cn
Journal of Software, [doi: 10.13328/j.cnki.jos.006631] http://www.jos.org.cn
©中国科学院软件研究所版权所有. Tel: +86-10-62562563
基于深度学习的新型视频分析系统综述
∗
孟令睿
1
,
丁光耀
1
,
徐
辰
1
,
钱卫宁
1
,
周傲英
1
1
(华东师范大学 数据科学与工程学院,上海 200062)
通讯作者: 徐辰, E-mail: cxu@dase.ecnu.edu.cn
摘 要: 摄像设备在生活中的普及使得视频数据快速增长,这些数据中蕴含丰富的信息.早期,研究人员基于传统
的计算机视觉技术开发视频分析系统,用于提取并分析视频数据.近年来,深度学习技术在人脸识别等领域取得了突
破性进展,基于深度学习的新型视频分析系统不断涌现.本文从应用、技术、系统等角度,综述了新型视频分析系统
的研究进展.首先,本文回顾视频分析系统的发展历史,指出了新型视频分析系统与传统视频分析系统的区别.其次,
本文分析了新型视频分析系统在计算和存储两方面面临的挑战,从视频数据的组织分布和视频分析的应用需求两
方面探讨了新型视频分析系统的影响因素.再次,本文将新型视频分析系统划分为针对计算优化的系统和针对存储
优化的系统两大类,选取其中典型的代表并介绍其核心设计理念.最后,本文从多个维度对比和分析了新型视频分析
系统,指出这些系统当前存在的问题,并据此展望了新型视频分析系统未来的研究与发展方向.
关键词: 视频分析系统;深度学习;计算优化;存储优化
中图法分类号: TP311
中文引用格式: 孟令睿,丁光耀, 徐辰, 钱卫宁, 周傲英. 基于深度学习的新型视频分析系统综述软件学报,2021,32(7).
http://www.jos.org.cn/1000-9825/0000.htm
英文引用格式: Meng LR, Ding GY, Xu C, Qian WN, Zhou AY. Survey of Novel Video Analysis Systems Based on Deep
Learning.Ruan Jian Xue Bao/Journal of Software, 2021 (in Chinese). http://www.jos.org.cn/1000-9825/0000.htm
Survey of Novel Video Analysis Systems Based on Deep Learning
MENG Ling-Rui
1
, DING Guang-Ya o
1
, XU Chen
1
, QIAN Wei-Ning
1
, ZHOU Ao-Ying
1
1
(School of Data Science and Engineering, East China Normal University, Shanghai 200062, China)
Abstract: The popularity of camera devices in our life has led to a rapid growth in video data, which contains rich information. Earlier,
researchers developed video analytics systems based on traditional Computer Vision techniques to extract and then analyse video data. In
recent years, deep learning has made breakthroughs in areas such as face recognition, and novel video analysis systems based on deep
learning have appeared. This paper presents an overview of the research progress of novel video analytics systems from the perspectives of
applications, technologies and systems. Firstly, this paper reviews the development history of video analytics systems and points out the
differences between novel video analytics systems and traditional video analytics systems. Secondly, this paper analyses the challenges of
the novel video analysis system in terms of both computation and storage, and discusses the influencing factors of the novel video analysis
system in terms of the organisation and distribution of video data and the application requirements of video analysis. This paper then
classifies the novel video analytics systems into two categories: those optimised for computation and those optimised for storage, selects
typical representatives of these systems and introduces their main ideas. Finally, this paper compares and analyses the novel video
analytics systems from multiple dimensions, points out the current problems of these systems, and looks at the future research and
development direction of novel video analytics systems accordingly.
Key words: video analysis systems; deep learning; computational optimisation; storage optimisation
随着大量的摄像机部署在公共场所甚至个人家庭中,这些设备产生的视频数据迅速增多.调查显示,2016 年
∗ 基金项目: 国家自然科学基金(61902128); 上海市扬帆计划(19YF1414200)
收稿时间: 2021-07-20; 修改时间: 2021-08-30; 采用时间: 2021-12-24; jos 在线出版时间: 2022-02-22
评论