BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Energy Minimization with Label Costs and Applications in Multi-Mod
 el Fitting - Boykov Yuri\, The University of Western Ontario
DTSTART:20110318T110000Z
DTEND:20110318T120000Z
UID:TALK30383@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:The a-expansion algorithm has had a significant impact in comp
 uter vision due to its generality\, effectiveness\, and speed. Recently\, 
 various extensions of a-expansion were proposed that in addition to data f
 idelity and spatial smoothness can optimize ``label costs'' with certain o
 ptimality guarantees. An energy with label costs can penalize a solution b
 ased on the set of labels that appear in it. The simplest special case is 
 to penalize the number of labels in the solution\, but the proposed energy
  is significantly more general than this. Usefulness of label costs is dem
 onstrated by a number of specific applications in vision that appeared in 
 the last couple of years.\n\nOur work (CVPR2010\, IJCV accepted) studies l
 abel costs from a general perspective\, including discussion of multiple a
 lgorithms\, optimality bounds\, extensions\, and fast special cases (e.g. 
 UFL heuristics). In this talk we focus on natural generic applications of 
 label costs is multi-model fitting and demonstrate several examples: homog
 raphy estimation\, rigid motion detection\, unsupervised image segmentatio
 n\, lossless and lossy compression\, and FMM. Our general approach is juxt
 aposed with classical K-means and EM. We also discuss a method for effecti
 ve exploration of the continuum of labels - an important practical obstacl
 e for discrete a-expansion algorithm in fitting models with continuous par
 ameters. We compare our optimization-based approach to multi-model fitting
  with standard extensions of RANSAC currently dominant in vision.
LOCATION:Small lecture theatre\, Microsoft Research Ltd\, 7 J J Thomson Av
 enue (Off Madingley Road)\, Cambridge
END:VEVENT
END:VCALENDAR