Pilot project in Edmonton aims to cut down on traffic casualties.

A new AI-powered traffic camera can identify distracted drivers on the road.

Developed by researchers at the University of Alberta in Canada, the camera was launched as part of a pilot project in south Edmonton.

Using photo radar technology, the large instrument resembles a light tower or power generator typically used on construction sites.

The AI model powering the cameras was trained on a sizable video dataset showing drivers using their mobiles and generally being distracted.

The system utilizes four distinct camera angles, looking into the driver’s windshield. A color context camera is used in the system. A third camera looks at license plates for vehicle identification. And a final camera identifies whether a phone is placed at the driver’s ear.

“It uses sensors to capture high-resolution images through windshields,” said Karim El-Basyouny, the urban traffic safety research chair at the University of Alberta.

The scientists received a grant to implement the technology in Canada. They’re working closely with the Edmonton Police Service to improve road safety. The city has a goal, dubbed “Vision Zero,” to have no traffic fatalities or serious injuries from road accidents.

The automated technology company, Acusensus, has implemented similar systems in Australia and India. The radar-based system doesn’t intrude on privacy or risk personal information, claims Tony Parino, Acusensus’s North American general manager. All the images undergo final inspection by humans to minimize errors.

The Canadian project is in the testing phase, with El-Basyouny suggesting it will “help us better understand what policies need to be in place in the future.”