Visual Tracking For Sports Applications
Speaker: Andrew Smith, ITEE
When: 2005-11-29 11:00:00
Venue: 78-420
Host: Prof Brian Lovell
Abstract:Visual tracking of human movement has attracted much attention due
to the wide variety of applications which could be performed
autonomously however currently need human interpretation. These
applications include sports training, rehabilitation and security.
Autonomous interpretation of human movement allows a much larger
volume of analyses to be performed at a much reduced cost.
Biometric analysis has already established itself as an effective
training tool for athletes, although most techniques rely on the use
of retro-reflective markers or magnetic sensors to be placed on an
athlete before such analysis can be performed.
Visual tracking of the human body involves the estimation of state
sequences with 30 or more dimensions. Due to high noise levels in
images the probability distribution associated with state estimation
is highly multi-modal. Particle filtering is a popular method used
for state estimation under such conditions. Many authors have
worked towards improving the efficiency of particle filters using
methods to concentrate particles in interesting regions of the
probability distribution. Currently these approaches are purely
data driven, which is problematic due to the high noise of the data.
In this work we seek methods to: improve the 'conditioning' of the
probability distribution, as well as improving the efficiency of
particle filters by using information specific to the problem of
visual tracking (e.g. camera geometry) and by working in a more
carefully chosen state space.
Biography:Andrew was born in Brisbane, Australia in 1979. He received a
Bachelor of Engineering(Electrical) from the University of
Queensland in 2001, and a Bachelor of Business Management(Economics)
also from the University of Queensland in 2002. He began work for
HTV on visual tracking projects and due to close collaboration with
Brian Lovell from the University of Queensland, enrolled in the
Masters of Philosophy program to give an academic framework for the
research he was conducting for HTV.
Type: Ph.D confirmation
Contact:Prof Brian Lovell, seminar host (lovell@itee.uq.edu.au)
or Guido Governatori (ITEE seminar co-ordinator)
(guido@itee.uq.edu.au)