Wednesday 5th October 2011, 13:40 - 14:30, FENS L058

Title : Sparse Occlusion Detection with Optical Flow and Its Application
       to Object Detection

Abstract:

We tackle two computer vision problems that are tightly coupled. The first one
is detection of occluded regions in a video stream with estimation of optical
flow simultaneously. Under the assumptions of Lambertian reflection and static
illumination, the task can be posed as a variational optimization problem, and
its solution is approximated using convex minimization. We describe efficient
numerical schemes that reach the global optimum of the relaxed cost functional,
for any number of independently moving objects, and any number of occlusion
layers.

Second, we describe a computationally efficient scheme segmenting a short image
sequence into an unknown number of objects. We exploit our work on occlusion
detection to bootstrap an energy minimization approach with depth ordering
constraints. The minimization problem can be converted into a linear program
with introduction of auxiliary variables and solved efficiently.The proposed
cost function integrates both appearance and motion features, and also achieves
model selection in a minimum-description length (MDL) setting.

Short bio:

Alper Ayvacı is a Ph.D. candidate at Computer Science Department in University
of California, Los Angeles under the supervision of Prof. Dr. Stefano Soatto.
His general research interests are in computer vision and applications of
geometry and convex optimization in this field. He is most interested in the
vision problems such as optical flow estimation, occlusion detection, image and
video segmentation, object recognition and super-resolution. He received his
B.Sc. and M.Sc. degrees in Computer Engineering at Istanbul Technical
University in 2003 and 2006, respectively.

Speaker : Alper Ayvacı, Wednesday 5th October 2011, 13:40 - 14:30, FENS L058