BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Stochastic geometry for automatic object detection and tracking - 
 Josiane Zerubia (INRIA Sophia Antipolis)
DTSTART:20171102T120000Z
DTEND:20171102T125000Z
UID:TALK94357@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:In this talk\, we combine the methods from probability theory 
 and stochastic geometry to put forward new solutions to the multiple objec
 t detection and tracking problem in high resolution remotely sensed image 
 sequences. First\, we present a spatial marked point process model to dete
 ct a pre-defined class of objects based on their visual and geometric char
 acteristics. Then\, we extend this model to the temporal domain and create
  a framework based on spatio-temporal marked point process models to joint
 ly detect and track multiple objects in image sequences. We propose the us
 e of simple parametric shapes to describe the appearance of these objects.
  We build new\, dedicated energy based models consisting of several terms 
 that take into account both the image evidence and physical constraints su
 ch as object dynamics\, track persistence and mutual exclusion. We constru
 ct a suitable optimization scheme that allows us to find strong local mini
 ma of the proposed highly non-convex energy. As the simulation of such mod
 els comes with a high computational cost\, we turn our attention to the re
 cent filter implementations for multiple objects tracking\, which are know
 n to be less computationally expensive. We propose a hybrid sampler by com
 bining the Kalman filter with the standard Reversible Jump MCMC. High perf
 ormance computing techniques are also used to increase the computational e
 fficiency of our method. We provide an analysis of the proposed framework.
  This analysis yields a very good detection and tracking performance at th
 e price of an increased complexity of the models. Tests have been conducte
 d both on high resolution satellite and UAV image sequences
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR