KBSU is testing a security system with a neural network that recognizes people’s poses

To improve the security system, the Kabardino-Balkarian State University is testing a neural network analytics system that is able to recognize different scenarios of people’s behavior in the lens of surveillance cameras. If a person raised their hands or lies on the floor, the university security service will receive an alarm signal. The RG correspondent found out how the system determines threats and in what direction such neural networks develop. The neural network pose detector, which «views» the image from the cameras, can respond to various scenarios of non-standard behavior and identify danger. At the same time, the AI that controls it is able to simultaneously work with hundreds of video sources simultaneously and without direct human participation. All this brings safety at the facility where such a system is installed to a new level.

Our university, whose educational buildings are located in several microdistricts of Nalchik, has a digital access control system (ACS), which ensures the anti-terrorist security of the university facilities. All entrance groups are guarded and they have a system of direct communication with the National Guard,

— said Yuri Altudov.

The data from the cameras is analyzed in a specially equipped center for monitoring and security management (the situational center of the KBGU), where specialists are on duty around the clock. They monitor what is happening at the university and respond to incoming signals from security posts and other security systems. According to Albert Shapsigov, Director of the Institute of AI and Digital Technologies, 520 IP cameras are currently used in the university.

They are installed in educational buildings, dormitory buildings and in the adjacent territory. The artificial intelligence systems being tested by the university’s specialists and developed by our partner, ITV Group, allow for computer analysis of events that fall into the field of view of video cameras,”

— Shapsigov noted.

The neural network analyzes the behavior of people 24/7 (when detectors are triggered for certain situations or scenarios: “shooting person”, “lying person”, “raised hands detector”. All these are potentially dangerous scenarios. For example, if one or several people pull hands in front of you as if holding a weapon, and other people raise their hands above their heads, such postures are a sign of an armed attack or robbery.If a person is lying on the floor or on the ground, this is a likely sign that he needs help.A quick reaction to a person falling in public places, construction and production facilities will also help to reduce the risk of negative consequences.Which situations and scenarios the system will react to is decided by those who operate it. the first results of testing neural network detectors at KBSU gave very encouraging results — the detectors correctly recognized the test situations and promptly sent alarms for the center dispatcher to work out. Testing of the AI system has been going on for several months, while artificial intelligence continues to learn already in a specific area.

Specialists of the Department of Computer Technologies and Information Security of the Institute of Artificial Intelligence and Digital Technologies, as well as the Center for Monitoring and Security Management (Situation Center) of KBSU work with the system. As the head of the Department of Computer Technologies and Information Security, Professor Tatiana Khashirova, said, in parallel with improving the security of the university, applied issues of using AI help in teaching students. These detectors are operational only if they are used correctly and the parameters are optimally adjusted. A direct line with a software developer helps us with this,” Khashirova explained. Based on the test results, KBSU will decide where and how to use the system in the future. These will be the most vulnerable and critical areas of the territory to improve the overall level of the university’s integrated security.