University develops AI tool to tackle knife crime

A new artificial intelligence (AI) system has been developed to help police identify knives and trace their origins more efficiently, researchers have said.

The system, Knife Hunter, has been created by Surrey University's Institute for People-Centred AI in collaboration with the Metropolitan Police.

The tool allows police to record weapons found, recovered or seized more quickly, and generate reports on knife crime patterns to support policing and inform policymaking.

The university described it as a "major step forward" with the potential to "create safer streets for everyone."

The system has been trialled as part of the Met's Operation Spectre which aims to tackle knife crime.

Prof Miroslaw Bober, the project lead at the Institute for People-Centred AI, said he hoped the system could be rolled out to other police forces and local authorities.

"For the past three years, we've partnered with the Metropolitan Police to develop a system we believe could revolutionise how forces across the country tackle knife crime," he said.

"It's a major step forward in helping uncover the origins of weapons in our communities.

"We're excited about the potential for Knife Hunter and similar tools to work with police and local communities to create safer streets for everyone."

More than 50,000 knife crime offences were recorded in England and Wales between June 2023 to June 2024 - a 4% rise from the previous year.

The Knife Hunter system has been trained on a data set of more than 25,000 images of 550 different knife types.

This means it can identify the slightest distinguishing features of a weapon, regardless of the viewing angle or lighting conditions, researchers said.

Prof Adrian Hilton, director of the Institute for People-Centred AI, said: "This research innovation is timely.

"We're showing how people-centred AI can transform lives and communities for the better."

Follow BBC Surrey on Facebook, on X. Send your story ideas to [email protected] or WhatsApp us on 08081 002250.