Event Location: NSH 1109
Bio: Alan Kraut is a student in the Robotics Institute, conducting research under David Wettergreen. His current research interests focus on the application of machine learning to computer vision for space robotics. He received a Bachelor of Science degree in Engineering from Harvey Mudd College in 2008.

Abstract: Data about Mars are being collected much faster than they can be analyzed by geologists for the creation of geologic maps. Because of this, automatic analysis of data would be very useful, even if it is less accurate than analysis performed by planetary scientists. A tool for analyzing orbital images would allow attention to be drawn to areas of Mars that are likely to be of particular interest. In this talk I will discuss a system for using data from multiple sources to classify areas of Mars terrain. Each scene was over segmented into superpixels, and a feature vector developed for each superpixel. We examined both a Bayesian classifier and boosting classifier using a bag of features approach, and a graphical model approach for incorporating spatial information into these results.