The detection of concealed firearm carrying trough CCTV: the role of affect recognition

This research aimed to explore whether the recognition of offenders with a concealed firearm by a human operator might be based on the recognition of affective (negative) state derived from non-verbal behaviour that is accessible from CCTV images. Since a firearm is concealed, it has been assumed that human observers would respond to subtle cues which individuals inherently produce whilst carrying a hidden firearm. These cues are believed to be reflected in the body language of those carrying firearms and might be apprehended by observers at a conscious or subconscious level. Another hypothesis is that the ability to recognize the carrier of concealed firearm in the CCTV footage might be affected by other factors, such as the skills in decoding an affective state of others and the viewpoint of observation of the surveillance targets. In order to give a theoretical and experimental basis for these hypotheses the first objective was to examine the extant literature to determine what is known about recognition of affect from non-verbal cues (e.g. facial expressions and body movement), and how it can be applied to the detection of human mal-intent. A second objective was to explore this subject in relation to the detection of concealed firearm carrying through performing a number of experimental studies. The studies employed experts, i.e. CCTV operators and mainly the lay people as participants. Also, various experimental techniques such as questionnaires and eye-tracking registration were used to investigate the topic. The results show that human observers seem to use visual indicators of affective state of surveillance targets to make a decision whether or not the individuals are carrying a concealed firearm. The most prominent cues were face, and upper body of surveillance targets, gait, posture and arm movements. The test of decoding ability did not show sufficient relationship with the ability to detect a concealed firearm bearer. The performance on the task might be view dependent. Further research into this topic will be needed to generate strategies that would support reliable detection of concealed firearm carrying through employing of related affective behavioural cues.