Researchers at MEPhI, a leading Moscow-based nuclear-focused university, have created special software to recognize people’s unusual behavior by the way they act in a crowd, including their manner of walking. The immediate market for the product may be an array of security systems, the MEPhI website announced.
To identify complex images by analyzing people’s walking style, stature, clothing style, etc., all the software needs are ordinary video recordings.
The software requires no additional costly equipment to operate, said Vadim Danshin, the project manager and a research fellow at MEPhI’s Institute of Smart Cybernetic Systems—using a regular smartphone is enough.
“We have plans for a range of tests to see how the product collects stats and analyzes customers and salespeople’s behavior in shopping malls. At later stages our algorithms can be used to identify unorthodox demeanor style recorded by surveillance cameras in airports, train stations or on public transport, an important task for modern IT-aided security systems,” Mr. Danshin added.
Other applications for the new software could include virtual reality and everything related to VR, interactive advertising in shopping malls, as well as a new family of simulators to train drivers, pilots, surgeons, the military or firefighters with automatic classification of their action.