Faster on the draw

Visual clues to detect if criminals are carrying guns are being studied to build a CCTV system that will help cut crime.

In an attempt to tackle gun crime in the UK, researchers from LoughboroughUniversity are developing an innovative identification system that will use CCTV cameras to spot individuals carrying concealed firearms.

Starting in June, the three-year multi-environment deployable universal software application (Medusa) project aims to develop intelligent software that can detect a person carrying a concealed weapon in real time.

While it is difficult to predict if someone is carrying a gun before crime occurs, Professor Alastair Gale, head of LoughboroughUniversity‘s Applied Vision Research Centre and leader of Medusa, said there are a number of cues the CCTV operator can pick up. These tend to be overt and covert cues (conscious and subconscious) and they will form the base of the intelligent software at the heart of the system.

The team will examine CCTV footage of people carrying concealed firearms to identify characteristics associated with the behaviour of criminals before they commit a gun-related crime. These will include body stance, gait, movement and eye contact with cameras. Once acquired, this information will be used to develop a novel machine-learning system for behavioural interpretation. Armed with this data, the CCTV cameras will scan footage automatically and match behavioural characteristics that indicate if an individual might be carrying a gun.

Researchers will use existing CCTV images and collaborate with the Royal Armoury at Leeds to generate a new set of images. These will cover a range of scenarios, covering armed and unarmed people and a variety of guns. ‘We are primarily looking into gun-carrying behaviour but the challenge is that it is difficult to obtain that much real-life footage of people carrying guns in the street,’ said Gale. He said the system will be developed to study knives as well, though at this early stage other weapons have not been considered.

In parallel with the project, researchers will meet experienced CCTV operators to establish the key cues they use to identify people carrying concealed weapons. ‘The idea is to marry human operator cues with the new software system which ultimately will produce an auditory and auto-visual cue to the operator,’ said Gale.

As a result of this work, the database of CCTV footage of individuals carrying or not carrying concealed firearms will be used by local authorities and the police for training CCTV operators to learn what to look for.

Gale added that it will be relatively straightforward to implement Medusa since the CCTV infrastructure is already in place.

It should also lead to more efficient use of police time and offer a more socially acceptable police practice for identification of potential criminals.

Gun crime is a growing problem in the UK – almost 11,000 firearm offences were recorded in England and Wales in 2004/2005, an increase of six per cent on the previous year.

The project, funded by a £620,000 grant from the Engineering and Physical Sciences Research Council, involves collaboration with Kingston, Brighton, Liverpool and SunderlandUniversities.