 |
| Credit: techworld.com |
A study has been conducted to develop an algorithm that would detect and recognise dangerous situations on CCTV, in particular, the presence of weapons such as knives and firearms.
The number of closed-circuit televisions (CCTVs) used for surveillance purposes in the United Kingdom has been widely discussed in past 10 years, ranging from 4.2 million to 1.85 million being the latest bet according to research undertaken by the ACPO lead on CCTV, Graeme Gerrard (NPCC, 2011). Forming a part of crime prevention strategy in the UK, and being often used as important evidence in court trials in the identification of suspects, the number of installed CCTVs seems now less important. Instead, the effectiveness of CCTV surveillance on crime reduction has been debated in recent years, raising some valid concerns.
In the study of CCTV cameras in north London, Sophie Watson, professor of sociology at the Open University concluded, that despite the crime statistics indicating decline in crime in CCTV occupied city areas, it did not have a lot of impact on perceptions of local residents. They did not feel secure, moreover, they felt that the presence of the CCTV cameras meant crime in their area was out of control. She also argued that CCTVs are not efficient enough in preventing crime, rather just moving it to another area with no cameras installed (Watson, 2008). On the other hand, the ‘Effects of Closed Circuit Television Surveillance on Crime’ review indicated that ‘CCTV has a modest but significant desirable effect on crime’, being most effective in reducing crime in car parks, when targeted at vehicle crimes. However, the review also concluded that ‘there was no effect of CCTVs on violence’ (Welsh, Farrington, 2008).
The results found that specificity of the knife detection algorithm was 94.93% and the sensitivity 81.18%, while for the firearm detection, the specificity was 96.69% and sensitivity 35.98%. A specificity of 100% was noted for the movie not containing dangerous objects. The algorithm generated no false alarms, making it a useful and valuable CCTV aid (Grega et al, 2016).
NPCC (2011). Graeme Gerrard: CCTV surveillance. Retrieved from http://www.npcc.police.uk/ThePoliceChiefsBlog/GraemeGerrardsCCTVblog.aspx.
Watson, S. (2008). Security in The City in Carter, S, Jordan, T. & Watson, S. (eds) Security: Sociology and Social Worlds, Manchester and New York, Manchester University Press in association with the Open University, pp.112 – 141.
The number of closed-circuit televisions (CCTVs) used for surveillance purposes in the United Kingdom has been widely discussed in past 10 years, ranging from 4.2 million to 1.85 million being the latest bet according to research undertaken by the ACPO lead on CCTV, Graeme Gerrard (NPCC, 2011). Forming a part of crime prevention strategy in the UK, and being often used as important evidence in court trials in the identification of suspects, the number of installed CCTVs seems now less important. Instead, the effectiveness of CCTV surveillance on crime reduction has been debated in recent years, raising some valid concerns.
In the study of CCTV cameras in north London, Sophie Watson, professor of sociology at the Open University concluded, that despite the crime statistics indicating decline in crime in CCTV occupied city areas, it did not have a lot of impact on perceptions of local residents. They did not feel secure, moreover, they felt that the presence of the CCTV cameras meant crime in their area was out of control. She also argued that CCTVs are not efficient enough in preventing crime, rather just moving it to another area with no cameras installed (Watson, 2008). On the other hand, the ‘Effects of Closed Circuit Television Surveillance on Crime’ review indicated that ‘CCTV has a modest but significant desirable effect on crime’, being most effective in reducing crime in car parks, when targeted at vehicle crimes. However, the review also concluded that ‘there was no effect of CCTVs on violence’ (Welsh, Farrington, 2008).
According to the latest police-recorded crime figures from the Office of National Statistics (ONS), there have been increases in murder and other violent crimes in England and Wales, including 13%-14% increases in gun and knife crime (ONS, 2017). The Metropolitan Police has launched intense proactive operations to tackle knife crime, as part of Phase Eight of their Operation Sceptre.
 |
| Credit: Metropolitan Police (May 2017) |
To use the CCTV surveillance more effectively, a study has been conducted to develop an algorithm that would detect and recognise dangerous situations on CCTV, in particular, the presence of weapons such as knives and firearms; as their detection and recognition could lead to more rapid response and a notable reduction in the number of casualties (Grega et al, 2016).
It has been argued, that even though human operators cannot be fully substituted by these automated alarms, they certainly can decrease the workload and help them stay focused, especially when a single operator monitors several cameras for hours, which can easily affect their judgement. To design the algorithm, several
facts had to be considered, such as that real-life CCTV recordings are usually
of inadequate quality, suffering from blurriness, compression and artefacts;
they are usually of low resolution due to poor quality; and that dangerous
object is visible only for a limited period of time in a scene, remaining
hidden by the perpetrator most of the time (Grega et al, 2016). Most
importantly, the aim was to keep the number of false alarms as low as possible,
as too many alarms could lead to being ignored by the operator, which would
make the whole system useless.
The results found that specificity of the knife detection algorithm was 94.93% and the sensitivity 81.18%, while for the firearm detection, the specificity was 96.69% and sensitivity 35.98%. A specificity of 100% was noted for the movie not containing dangerous objects. The algorithm generated no false alarms, making it a useful and valuable CCTV aid (Grega et al, 2016).
Grega et al. plan to continue their work on the algorithms in order to provide
a complete and ready-to-market solution for CCTV operators. They also ‘plan to integrate both
algorithms into a single solution, while further focusing on reducing false
alarms and increasing sensitivity’.
References:
Grega et al. (2016). Automated Detection of Firearms and
Knives in a CCTV Image. Sensors
16 (1), 47. DOI:
10.3390/s16010047.
NPCC (2011). Graeme Gerrard: CCTV surveillance. Retrieved from http://www.npcc.police.uk/ThePoliceChiefsBlog/GraemeGerrardsCCTVblog.aspx.
ONS (2017). Crime in England and Wales: year ending
Dec 2016. Retrieved from https://www.ons.gov.uk/peoplepopulationandcommunity/crimeandjustice/bulletins/crimeinenglandandwales/yearendingdec2016.
Watson, S. (2008). Security in The City in Carter, S, Jordan, T. & Watson, S. (eds) Security: Sociology and Social Worlds, Manchester and New York, Manchester University Press in association with the Open University, pp.112 – 141.
Welsh, B.C. and Farrington, D.P. (2008). Effects of Closed
Circuit Television Surveillance on Crime. Retrieved from https://www.campbellcollaboration.org/media/k2/attachments/1048_R.pdf.