Can CCTV reliably detect gun crime?
online resourceposted on 2008-04-22, 11:17 authored by Iain T. Darker, Alastair Gale, Leila Ward, Anastassia Blechko
In recent years the highest levels of gun crime in England and Wales have been recorded. Gun crime is known to occur in environments that fall under the view of the networks of CCTV cameras which are endemic in urban Britain. Consequently, CCTV has the potential to detect gun crime. However, CCTV operators face a near impossible task: often, each individual operator is required to monitor multiple cameras for extended time periods. It is not surprising then to find that CCTV has had little effect at reducing gun crime. A new collaborative research programme (MEDUSA) is underway across the UK which will develop a software application to aid CCTV operators detect the precursors of gun crime. MEDUSA aims automatically to detect both guns as objects (concealed or otherwise) and people who intend to commit gun crime, via CCTV. The system is partly based on eliciting the features that skilled CCTV operators identify as indicative of a concealed weapon or suspicious behaviour related to gun crime. Initial studies into the examination of these skills across several CCTV centres indicate that there was little formal instruction in the detection of mal-intent. Much skill is built up from prolonged experience in examining CCTV imagery. Thus, potentially, there is widespread individualisation of detection skills. Further, visual cues specific to gun crime were limited. However, the present methodology relied upon direct report. It is argued that a range of both consciously and subconsciously detected cues could facilitate gun detection via CCTV. Future studies will examine this possibility.
- Computer Science