An Unsupervised Approach to Anonymous Crowd Monitoring
Ian Hales, Roger Boyle, Kia Ng · 2010
With over 4.2 million CCTV cameras in the UK alone [2], it would be useful haveto an automated system to monitor over wide, open spaces. In such areas, we canobserve emergent behaviour in crowd movements, often changing dynamicall over timeas crowds form and disperse. Trained operators can often notice trouble the moment,if not before, it happens. Unfortunately, the high number of cameras watching over thepublic has generated a feeling of unease within the populus as people feel increasinglythat their privacy is being invaded.We propose a system that, using an offline, unsupervised lear ning process, willanonymously detect patterns of motion within a scene and describe it as usual or un-usual. The system is trained on footage of the scene, recorded using a single camera.Flow is detected using the KLT tracker [3] to accumulate, at chosen granularity, ‘track-lets’ [1] of elemental motion.These tracks are quantised and their distribution in spatial and temporal windowsaround each pixel clustered to generate acceptable patterns. In scenes with changingbehaviour, there may be several candidates at each postion. During testing, similarlygenerated patterns are tested for plausibility by proximity to acceptable clusters.Early results show promise and may be tuned via various parameters of the system.