近年来,随着视频监控系统的大量普及安装,监控系统所获取并存储的视频数据容量正以惊人的速度增长。从理想角度看,这些视频包含了现实世界中的大量信息,应该为我们的管理及安保工作带来巨大的价值。但是从现实角度看,依靠人工处理包含数以万计的视频数据集,并从中获取信息,是非常困 In recent years, with the widespread adoption of video surveillance systems, the amount of video data captured and stored by the surveillance system is increasing at an alarming rate. Ideally, these videos contain a wealth of information in the real world and should bring great value to our management and security efforts. But from a realistic point of view, it can be very difficult to rely on manual processing of information that contains tens of thousands of video datasets