In the present-day context, security has emerged as a matter of utmost significance, particularly for individuals and businesses striving to safeguard their assets and people. While conventional video surveillance systems have long been relied upon, they come with inherent limitations.
Machine learning, covering today various areas of human activity, is used in a classical form for training on a large amount of data. The algorithms of most systems implemented on the basis of a personal computer and united by the general term motion analysis are based on methods for analyzing optical flows or correlation motion. Software works most accurately when a person is directly in front of the camera, selects moving objects and monitors them, detecting the appearance of new ones or the disappearance of previously observed ones.