VASC Seminar Announcement
        =========================

 Speaker:   Stella X. Yu
 Affiliation:  Carnegie Mellon University, Robotics Institute and CNBC
 Date: Monday, September 30
 Time: 3:30pm
 Place: NSH 1507 

 Title:  Segmentation through Data-Model Interaction

  Abstract:
  Segmentation and recognition have long been treated as two separated
  processes.  We propose a mechanism based on spectral graph
  partitioning that readily couple the two processes into one.
  A part-based object recognition system detects patches, supplies their
  partial segmentations and an evaluation of patch labelling consistency.
  A pixel-based image segmentation system detects edges and evaluates
  feature similarity between pixels.  We model the coupling between the two
  process in the solution space and show that the joint optimization can be
  done through eigendecomposition.  We demonstrate on real images that
  object part consistency and image pixel coherence
  can work together to eliminate false positives and labelling
  ambiguities at high level, while at the same time overcome occlusion
  and weak contours at low level.