Retina to Robots
報告人:Dr Dermot Kerr, Ulster University, U.K.
報告時間:2017-6-19 9:00-11:00 、6-21 9:00-11:20
報告地點:信息學館301、信息學館224
Dermot Kerr received a B.Sc. (Hons) in Computing Science from the University of Ulster, UK in 2005, and a Ph.D. in Computing and Engineering from the University of Ulster, UK in 2008. He is currently a lecturer in the School of Computing and Intelligent System at the University of Ulster, Magee. His current research interests are in biologically inspired image processing, mathematical image processing, feature detection, omni-directional vision, system identification and robotics. Dr. Kerr is an officer and member of the Irish Pattern Recognition and Classification Society. He is currently involved in the EU FP7 funded projects VISUALISE and SLANDAIL.
Abstract
The processing capabilities of biological visual systems are still vastly superior in terms of performance for real-time and low-power applications compared with conventional artificial vision. There is increasing evidence that biology has evolved a multitude of cell types, including at the level of the retina, to adapt to an extensive set of dynamic visual environments. Existing bio-inspired artificial vision technology has failed to consider the utility of modelling this rich diversity of cells, despite the fact that these cells are crucial to biology’s ability to process the natural visual environment.
This talk will demonstrate that, by creating a refined understanding of retinal function in natural visual environments, enhanced computational models which emulate the biological signal processing in the retina can be used for many challenging high-speed robotic and computer vision applications.