Cameras Spot 19,000 Potential Driving Offences

Potential Driving Offences

Cameras Spot 19,000 Potential Driving Offences in Just Two Weeks

A road safety trial in the West Midlands has identified over 19,000 potential driving offences in just two weeks using new artificial intelligence (AI) cameras.

The images of potential driving offences include drivers using their mobile phone and passengers failing to wear seatbelts.

Among the images captured is a front-seat passenger with a small child on their lap with no seatbelt, leaving the toddler completely unprotected in the event of a collision. It is one of dozens of images of children being driven without age-appropriate restraints.

In another case a driver was captured holding their mobile phone with both hands, eyes on the screen rather than the road.

The RAC reports that during the two-week trial, funded by the West Midlands Combined Authority, was conducted in March using advanced AI camera systems positioned outside the Ladywood and Fallings Park fire stations in Birmingham and Wolverhampton.

Initial scans of the images identified 11,144 potential seatbelt offences and 1,163 potential mobile phone offences at the Ladywood site.

At Fallings Park, the cameras detected a further 6,081 potential seatbelt offences and 757 potential instances of drivers using a mobile phone while behind the wheel.

Mat MacDonald, West Midlands Road Safety Commissioner said: “Mobile phone use at the wheel and failing to wear a seatbelt are two of the leading causes of death and serious injury on our road network, and yet you barely have to leave your front door before you see someone taking these potentially lethal risks.

“Our work with this cutting-edge technology shows we will stop at nothing to stamp out such behaviours, fostering a safer culture of driving on our streets and preventing yet more families going through the anguish of losing a loved one to avoidable tragedy.”

The AI camera technology was developed and supplied by AECOM/Acusensus, with its accuracy checked by human operators using a sample of 8,024 images.

You might also like

More Blog Posts