TITLE: Traffic Density Estimation with On-line SVM Classifier

PRESENTER: Zhidong LI, http://www.cse.unsw.edu.au/~zlix679/, zlix679@cse.unsw.edu.au

AFFILIATION:School of Computer Science and Engineering, UNSW; NICTA, http://www.cse.unsw.edu.au

DATE: Tuesday 28th July 2009

TIME: 11:30:00

PLACE: CSE Seminar Room, Level 1, K17

ABSTRACT:

Determining the vehicular traffic density in an intelligent transport
system (ITS) is presently attained mainly through loop detectors (LD),
traffic radars and surveillance cameras. However, the difficulties and
cost of installing loop detectors and traffic radars tend to be
significant. Currently, a more advance method of circumventing this is to
develop a sort of virtual loop detector (VLD) by using video content
understanding technology to simulate behavior of a loop detector and to
further estimate the traffic flow from a surveillance camera. Such a
virtual loop detector requires supervised training with certain human-aid
efforts in its setup. The difficulties also arise when attempting to
obtain a reliable and real-time VLD under changing illumination, weather
conditions and static shadows. In this paper, we study the effectiveness
of using texture feature in describing the traffic density, and propose a
real-time VLD via an on-line SVM classifier together with background
modeling technique (OSVM-BG) to estimate the traffic density states
probabilistically and automatically. The system uses the feedback from the
background modeling to train and update its SVM kernel so as to adapt
itself to the difficult lighting environment. From the testing we find
that the system outperforms several existing algorithms and achieves an
average accuracy at around 90% under various illumination changes, weather
conditions and especially changing static shadows in daytime. This work
will be presented at AVSS this year.

BIOGRAPHY OF SPEAKER:

Zhidong Li received the B.S. degree from Xiamen University, China, in 2002,
and the M.S. degree from the University of New South Wales, Australia, in
2006. He is currently a PhD student in the School of Computer Science and
Engineering at the University of New South Wales and National ICT Australia.
Zhidong is a student member of the IEEE. His research interests include
image processing, computer vision, machine learning, and statistical pattern
recognition.

Host:

Yang Wang

Seminar Convenor:

Van Hai Ho