Visual tracking, a building block for many applications, has challenges such as
occlusions, illumination changes, background clutter and variable motion
dynamics that may degrade the tracking performance and are likely to cause
failures. In this presentation, Track-Evaluate-Correct (self-correlation)
framework will be discussed in order to achieve a robust tracking over existing
trackers. For existing trackers, the framework integrates an evaluation unit to
check the status of tracking quality online and a correction unit to avoid
upcoming failures or to recover from failures. A Dynamic Bayesian Network (DBN)
model will be described for a generic representation of the self-correcting
tracking and particular trackers developed using the Track-Evaluate-Correct
framework will be discussed.​

Tewodros A. Biresaw is a PhD student at Queen Mary University of London - UK,
and University of Genova - Italy, under the supervision of Prof. Andrea
Cavallaro and Prof. Carlo S. Regazzoni, respectively. He received double M.Sc.
degree in European Master on Advanced Robotics (EMARO), from Warsaw University
of Technology, Poland, and University of Geneva, Italy, in 2010, and B.Sc.
degree in Electrical Engineering from Bahir Dar University, Ethiopia, in 2005.
His research interests include self-correcting tracking, self-aware systems,
data fusion, Probabilistic Graphical Models and computer vision.

​