ADABTS work on “Abnormal Behaviour Definition” informs the development of an innovative system to enable automatic detection of unusual or suspicious behaviour in crowded spaces.
European scientists are developing a futuristic system for the automatic detection of abnormal potentially threatening behaviour that might potentially be used to protect large crowds at sporting events (such as the Olympic Games), in town centres or at airports. The project’s goal is to facilitate the protection of EU citizens, property and infrastructure against threats of terrorism, crime, and riots, by early detection and alerting of potential threats to CCTV operators and other security staff.
The four year €3.5m project, called ADABTS or Automatic Detection of Abnormal Behaviour and Threats in crowded Spaces, was launched in August 2009, and is jointly funded by the European Commission and the partners involved who form an international consortium consisting of FOI (SE), BAE Systems (UK), Detec (NO), Centre for Applied Science and Technology (UK), Institute of Psychology – Ministry of the Interior (BG), SINTEF (NO), University of Amsterdam (NL) and TNO (NL).
ADABTS draws on past security incidents, which have demonstrated that crowded spaces are vulnerable to terrorist attacks and criminal activity. There is therefore a real need to identify hostile or threatening activity in crowded spaces especially prior to an incident or an attack.
The project has recently produced its second report entitled “Abnormal Behaviour Definition”. This report describes a study that set out to determine what features a human operator subconsciously found suspicious within CCTV surveillance images and what features might be used to inform judgements and decisions about the occurrence of abnormal behaviour.
Focusing on large-scale events taking place in sports arenas, crowded public spaces and critical infrastructure, the study defined a series of indicators of abnormal behaviour for each context. The indicators were drawn from previous research, analysis of past incidents, by using eye tracking techniques in conjunction with post-event interviews whilst operators viewed CCTV images to extract tacit knowledge and finally a survey of domain experts. Distinct and visible behaviour, such as whole-body behaviours (including movement about a space, excessive body gestures or gait), were identified as well as behaviours that are less obvious (such as signs of stress, eye movements, mumbling and sweating). Appearance indicators and audio indicators were also considered.
The study then modelled these indicators and surveyed the domain experts to determine what behavioural indicators are important in detecting abnormal behaviour and patterns of movement in crowded public spaces. The behaviour indicators rated as important were relatively constant across the contexts, however the ratings for their level of utility varied, indicating that it would be more useful to detect some more accurately and consistently than others.
Preliminary tests on the ADABTS system have been started and will continue through 2012 based at a football stadium (ADO Den Haag) in the Netherlands. The tests employed professional actors playing out “abnormal behaviours” and “scenarios” which had been identified by the project partners as important for the ADABTS system to detect. A final demonstration of the ADABTS system is planned at ADO Den Haag in 2013.
***The “Abnormal Behaviour Definition” report and further information regarding the project is available at the ADABTS project site at http://www.adabts-fp7.eu.