Artificial intelligence safety cameras save lives
An academic from Nottingham Trent University has developed artificial intelligence safety cameras. Capable of detecting dangerous levels of crowding with one hundred per cent accuracy during day or night, the smart camera system aims to help save lives by improving crowd safety at large open space public gatherings.
Designed by a research team lead by Dr Amin Al-Habaibeh of the School of Architecture, Design and the Built Environment, the new technology can also give a strong estimation of crowd numbers. Combining an infrared camera with a monochrome visual camera, the artificial intelligence safety cameras measure the real time density of crowds and signals when and where hazardous levels of overcrowding occur in various spots.
Dr Al-Habaibeh, who carried out the research in conjunction with the Um Al-Qura University in Mecca, Saudi Arabia, commented: “Large scale public events still present huge safety risks and there are several recent examples of incidents which have resulted in loss of life. It’s important that technology helps minimise those risks and, where possible, helps prevent any instances such as crushing. The project has many potential applications around the world including sport events and city festivals. This invention is a step forward as it will provide real time data during day or night, in fog or smoke, about how dense a crowd is in an open space. It will enable those in charge of safety to take swift action before any sort of incident occurs.”
With the two types of cameras linked to a computer with temperature and light sensors, tests revealed that the artificial intelligent system provided a 0% margin of error for detecting different crowd densities. The monochrome visual camera detects the outline of individuals in the crowd, while the infrared camera senses the heat emitted from their bodies.
"The ability of the system to operate in the dark or smoke makes it particularly useful for emergency situations. It will give security staff the knowledge of whether or not people’s safety is at risk, even when this is impossible tell with the naked eye,” comments Saied Yaseen, Nottingham Trent University’s postgraduate researcher on the project.
The artificial intelligence system calibrates the side-effects of sunlight and of people in the foreground appearing larger to the camera than those in the background. Considered to be extremely high, an alert signal can be activated when a density of five persons or more per metre² occurs.