小视频内容分析技术发展探讨  

Short Video Content Analysis Technology

作　　者：薛向阳[1] 李斌 XUE Xiangyang;LI Bin(Fudan University,Shanghai 200433,China)

机构地区：[1]复旦大学计算机科学技术学院,中国上海200433

出　　处：《中兴通讯技术》2021年第1期54-59,共6页ZTE Technology Journal

摘　　要：小视频数量呈爆炸式增长态势,并引发了许多技术需求,包括小视频的编辑、搜索、推荐、溯源、审查和监管等。介绍了小视频数据的主要特点和小视频内容分析技术面临的挑战,并对目标检测、追踪、重识别等小视频内容分析技术的研究进展做了综合分析。认为只有构建一个整合多种不同算法的系统,才能够更准确、更鲁棒地解决分析问题,才能系统性地完成小视频内容分析任务。The number of short videos has increased explosively,which has led to more technical requirements,such as editing,searching,recommendation,sourcing,censoring,and monitoring of short videos.The main features of short video data and the challenges faced by the short video content analysis technology are introduced.The research progress of short video content analysis technologies such as object detection,tracking,and re-identification is comprehensively analyzed.It is considered that only by building a system that integrates multiple algorithms,can the analysis problems be solved more accurately and robustly,and the short video content analysis task can be completed systematically.

关 键 词：小视频 内容分析技术 视频目标检测 多目标追踪 人物重识别 

分 类 号：TN919.8[电子电信—通信与信息系统]