Answers from Bosh.

«Our security company services a shopping center. We often have to process large volumes of video information in search of a specific event. How can this process be accelerated? — Igor Moseychenok, Ukraine» .

Sergey KUZNETSOV, Director of Distribution, Technical Support and Marketing for Bosch Security Systems Russia – Ukraine – Belarus – Kazakhstan, answers the reader’s question.

To solve this problem, it is necessary to use a modern intelligent video data analysis system that works independently on each camera and does not require a central analytics server. It should have a wide range of detection functions, from stationary object tracking to motion trajectory tracking. In such systems, real-time images are instantly analyzed, and the resulting data stream accompanies the video stream. All events are displayed immediately, and the data can be saved for later use. The system should be a comprehensive solution and record all the details, giving the user the ability to search for any event, even one that was not configured to generate an alarm, storing data on everything that happens in the active areas of each monitored scene.
The system generates analytical information and stores it as metadata along with the video image. Metadata contains detailed information about all objects that are in the monitored area and moving in and out of it. The analysis is not limited to real-time images: the system provides event recognition when playing back recorded video images. Recorded metadata, consisting of simple text strings describing specific image details (such as size, shape, color, speed and direction of movement, or the outline of a person's head), is much smaller and easier to search than recorded video images. By searching metadata using intelligent functions similar to those of Internet search engines, the system instantly takes the user to the relevant part of the video without scanning the entire video stream.