Scientists learn to detect unusual human behavior by gait

Specialists of the National Research Nuclear University "MEPhI" (Russia) have created a program that recognizes unusual behavior of people according to their behavior in the crowd, including the manner of walk. This development can find application in security systems.

The program recognizes complex images via usual video. The algorithm requires implementing only two conditions – the camera must remain stationary, and the color of the clothes should not match the background. At this stage, scientists are completing work on the program to recognize people in the crowd by the manner of walk, body type, clothing style, etc. With the support of the MEPhI Engineering center they are preparing to test program opportunities and looking for commercial partners.

According to the project manager, employee of the MEPhI Institute of cyber intelligent systems Vadim Danshin, the program does not require additional expensive equipment, the power of a regular smartphone is enough.

"We plan to conduct tests of this development on gathering statistics and analysis of buyers and sellers behavior in shopping malls. In the future, our algorithms can be used to detect unusual behavior recorded by the surveillance cameras in airports, stations and transport. This is the challenge for modern IT-security systems," said Vadim Danshin.

The development can also become a virtual reality game and all that is connected with it, interactive advertising in shopping centers, creation of simulators for drivers, pilots, surgeons, soldiers and firefighters with automatic classification of their actions.

Provided by National Research Nuclear University