With random terror attacks and other forms of hate crimes on the rise worldwide, authorities are hard pressed to find a solution to save innocent lives. While police and secret service agencies are continuously upping monitoring, scientists have come up with another solution. They have devised a method to keep a tab on the movements of hundreds of individuals at the same time and uncover suspicious behavior.
What differentiates the software from others so far and how does it work?
The research developed by Mubarak Shah and Afshin Dehghan and to be published in IEEE Transactions on Pattern Analysis and Machine Intelligence, leverages a mathematical function to study five factors that will aid in finding out where a person will go next. The software registers appearance, neighbor motion, target motion, spatial proximity, and grouping to find out how an individual will move around in a crowd and even predict their likely actions.
The developers of the algorithm studied nine crowd videos, in which the number of people ranged from 57 to 747. The program’s effectiveness in tracking every person was between 67% and 99%. The program is still being tested before it can be launched in the market. Once it happens, it shall hopefully go a long way in thwarting untoward incidents in crowded places.
So far, computer-based studies have only been able to monitor one person at a time in recorded videos and mostly rely on appearance. The latest algorithm, however, is slated to change all that by enabling surveying the entire crowd at a time.