Colloquium "Recognizing 3D objects using temporal ensembles of shape functions"

Abstract:

3D object recognition is a recurring problem in many heterogeneous domains, such as computer graphics, computer vision, robotics etc. This talk will introduce Temporal Ensembles of Shape Functions, novel descriptors that are useful in compactly representing multiple partial views of a 3D object. Target application scenario involves a mobile robot observing an object with an RGB-D camera from a limited number of views. Preliminary experimental results with synthetically rendered objects will be presented, using two well-known benchmark datasets: the 3D-Net database and the Princeton Shape Benchmark.

CV:

Karla Brkić received her M. Eng. (dipl. ing.) degree in Computing in 2007, and her PhD in Computing in 2013, all from the Faculty of Electrical Engineering and Computing, University of Zagreb. Since 2008, she has been employed at the same Faculty, as a research associate at the Department of Electronics, Microelectronics, Computer and Intelligent Systems. She participated in various international research projects, and has been extensively collaborating with the Vision-based Measurement Group of the Graz University of Technology and the Vision4Robotics Group of the Vienna University of Technology, Austria. Since September 2012, she participates in the ACROSS project, where she is a member of the Applied Cognition and Vision Group. She has co-authored four journal papers and thirteen papers published in the proceedings of international conferences. Her research interests include representation, classification and reasoning about data in computer vision.