| ECVision indexed and annotated bibliography of cognitive computer vision publications This bibliography was created by Hilary Buxton and Benoit Gaillard, University of Sussex, as part of ECVision Specific Action 8-1 The complete text version of this BibTeX file is available here: ECVision_bibliography.bib
T. Troscianko and A. Holmes and J. Stillman and M. Mirmehdi and D. B. Wright
ABSTRACT
Can closed-circuit TV (CCTV) surveillance allow potentially antisocial or criminal behaviour to be predicted? We report a study whose aims were (a) whether observers can successfully predict the onset of such behaviour; (b) whether there may be a difference between naive and professional observers; (c) where, in the sequence of events, it is possible to make this prediction. We obtained 100 sample scenes from urban locations in UK. 18 of these led to criminal behaviour (fights, vandalism). A further 18 scenes were matched as closely as possible to the crime example s, but did not lead to any crime. 64 were neutral scenes chosen from a wide variety of non-criminal situations. A signal-detection paradigm (yes/no, and 6-point confidence rating scale) was used. Results on 50 naive and 50 professional observers show (a) that observers can distinguish crime sequences from matches, with d'e=1.17; (b ) that there is no difference between the naive observers and the experts; (c) that there are key types of behaviour (particularly gestures and body position) that allow prediction to be made. In principle, it may be possible to detect such aspects of behaviour and therefore produce alerting mechanisms. |
|
| Site generated on Friday, 06 January 2006
|