Cornelia Fermuller
University of Maryland
Research Scientist
Action Perception at Multiple Time Scales
Sensory Anticipation in Action Understanding
Object Motion Segmentation: Advantages from Event Data
Optical Illusions in Man and Machine
Email hidden; Javascript is required.
My research has been on the integration of perception with action, with a focus on biologically inspired solutions. I have published in the areas of Robotics, Computer Vision, and Human Vision. I have modeled problems and developed software in the areas of multiple view geometry, navigation, motion, stereo, shape, texture, action and object recognition, and image processing. I have also combined computational modeling with psychophysical experiments to gain insight into human motion perception and low-level vision. My current research focuses on two themes. 1) Developing cognitive systems that understand and can learn human actions, through the integration of language and reasoning with perception and action 2.) Motion tasks in extreme conditions. The goal in this task is to study and develop algorithms on data from event-based bio-inspired sensors, which hold promise for challenging applications in autonomous robotics, because of their high-dynamic range and low latency.