The system analyses data from security cameras with "80 percent accuracy rate (which) is approximately 30 per cent higher than that of conventional methods" which use historical data on rates of flow of crowds involving highly complex computations and are not sufficiently accurate, a Mitsubishi spokesperson told EFE news.
The technology, developed together with the Research Centre for Advanced Science and Technology of Tokyo University uses a movement simulator in real time, which reduces computational complexity as compared to existing methods through a high-speed human-behaviour model, the company said.
The system can provide support to event organisers as an estimate in real-time will allow them to put in place preventive crowd management systems, said Mitsubishi Electric.
Mitsubishi will test the system on August 20, during the fireworks celebrations on the Tama river west of Tokyo -- where a large gathering is expected -- and plans to market the system as soon as possible, although it does not yet have a specific launch date, added the spokesperson.
