Video surveillance systems produce massive amounts of video. Due to lack of time or resources however, most of this video never gets watched or reviewed. As a result, security incidents get missed and suspicious behavior is not detected in time to prevent incidents. These challenges have played a large part in the development of video analytics.
Video analytics are software applications that automatically generate descriptions of what is actually happening in the video (so-called metadata), which can be used to list persons, cars and other objects detected in the video stream, as well as their appearance and movements. This information can then be used as the basis on which to perform actions, e.g. to decide if security staff should be notified, or if a recording should be started.
Figure 1: An example of video analytics: When a person crosses a virtual line (defined in the analytics application in the network camera), metadata from the event is generated and a notification is sent to a PC.