Benutzerspezifische Werkzeuge

Machines that see (In Zusammenarbeit mit dem Institut für Klassische Archäologie der LMU)

Seeing is apparently done effortlessly by the human and animal brains, but what really lies behind the interpretation of this most complex stream of sensations? There are several ways to attack this question. One approach is to prod and probe the living brain in action, or to observe it through advanced medical imaging instruments. Another is to record the psychological responses of human and animal subjects to visual stimuli. This lecture takes a third approach – looking at what might be required to simulate the intelligent behaviours that make up the ability to see. Taking inspiration from psychology and illusion, the notion is explored that seeing is an inherently chancy business. In common with other tasks that involve intelligence, it is argued that seeing is done by forming hypotheses about what is out in the world. It is a matter of continually weighing and re-weighing possible interpretations, in the light of incoming evidence. At the core of this activity is reasoning with probabilities. These ideas will be illustrated in the lecture using a variety of demonstrations of artificial vision systems.

Prof. Andrew Blake
has been an academic for most of his professional life, first in the University of Edinburgh, then (1987-99) as faculty and Professor at the University of Oxford, where he also held a Royal Society Senior Fellowship. Recently he moved to the European laboratory of Microsoft Research in Cambridge, England as Senior Researcher, where he continues to study Artificial Intelligence, especially with respect to Machine Vision.