Course DescriptionLectures: , Tue 4:30-6:00 SC 220, Thu 4:30-6:00 SC 221
This a seminar course that examines the classical statistical approaches to visual motion analysis. Several areas of visual motion will be examined like 3-D structure and motion estimation, optical flow, segmentation and tracking. A variety of statistical techniques will be presented with emphasis on how they can be applied on vision problems. These techniques include Least Squares, Clustering, Kalman Filter, Maximum Likelihood, Maximum A Posteriori, Expectation Maximization, Robust Methods etc. Approximately half the lectures will be on background theory by the instructor and the other half literature review by the students.
EvaluationEvery student will write 3 papers, two of them short literature reviews, that cover a rather narrow area, are based on several papers and are presented in class while the third (last) is much longer, and can be either a literature review or an original but well contained research topic. The short papers carry a weight of 30% and the long 40%. All presentations are 20 minutes.
Last Modified: Jan 8 2013