Human Action Recognition
Human Action Recognition
复制标题
人类行为识别
DOI：
10.1201/9781439813300-c8
复制
发表时间：
2009
期刊：
影响因子：
--
通讯作者：
Baoxin Li
中科院分区：
文献类型：
--
作者：
X. Xu;Baoxin Li
随着公共区域越来越多地安装监控摄像头，从银行和超市的小规模独立应用到公共街道执法应用中的大规模网络闭路电视（CCTV），对计算机视觉技术的需求越来越大，以了解现场的人类活动。这是计算机视觉领域一个极具挑战性的研究热点。本次演讲将介绍我们正在进行的人类行为识别项目的最新成果。该方法包括三个主要步骤，即人体运动检测，人体运动特征提取和人体动作分类器。流行的公开数据库，包括KTH，魏茨曼，CAVIAR和PETS的结果，将报告。
With the growing installation of surveillance cameras in public areas, ranging small-scale stand-alone applications in banks and supermarkets to large-scale networked-close-circuit television (CCTV) in law enforcement applications in public streets, there is an increasing demand of computer vision technology for understanding human activity in the scene. This is a very challenging and hot research topic in computer vision community. This talk will present our recent results of our ongoing project on human action recognition. The proposed method consists of three major steps, namely human motion detection, human motion feature extraction and human action classifier. Results on popular publicly available databases including KTH, Weizmann, CAVIAR and PETS, will be reported.
影响因子：
19.5
作者：
Niebles, Juan Carlos;Wang, Hongcheng;Fei-Fei, Li
通讯作者：
Fei-Fei, Li