Head Tracking and Action Recognition in a Smart Meeting Room
Hammadi Nait‐Charif, Stephen James McKenna · Discovery Research Portal (University of Dundee) · 2003
Computervision-based monitoring was used for automated recognition of the activities of participants in a meeting. A head tracker, originally developed for monitoring in a home environment, was evaluated for this smart meeting application using the PETS-ICVS2003 video data sets. The shape of each person’s head was modelled as approximately elliptical whilst internal appearance was modelled using colour histograms. Gradient and colour cues were combined to provide measurements for tracking using particle filter s. A particle filter based on Iterated Likelihood Weighting (ILW) was used which, in conjunctionwith the broad likelihood responses obtained, achieved accurate tracking even when the motion model was poor. It was compared to the widely used Sampling Importance Resampling (SIR) algorithm. Results are reported for tracking and recognition of the actions of the six meeting participants in the PETS-ICVS data. ILW outperformedstandard SIR and reliably tracked all participants throughout the meeting scenarios. A head tracker has been developed for use in a home monitoring application which aims to support elderly people to live independently. This paper reports an evaluation of the application of this tracker to a smart meeting room using the PETS-ICVS data sets. The aim was to automatically annotate the activities of the meeting participants based on video data. Our premise was that reliable tracking of the head of each person would yield interesting annotation data in terms of motion trajectories and that these in turn could be used to recognise certain actions such as standing up, sitting down, entering, exiting and walking to the whiteboard. The system needed to be able to simultaneously track multiple people, perform automatic initialisation, handle person-person oc