View-based representations in animal and machine vision

View-based representations in animal and machine vision

This meeting will bring together experts in computer and biological vision who are interested in possible forms of spatial representation, particularly representations that avoid the need for 3D reconstruction. The focus will be on alternative view-based methods.

In animals from insects to man, internal representations enable navigation, spatial judgements and object recognition in a 3D world. Computer vision has shown how to build an explicit 3D map from visual information, but using biologically implausible architectures. However, in computer vision, topological or 'view-based' representations are increasingly replacing 3D reconstruction. Could similar 'view-based' representations help to explain biological vision?

Meeting Information

Organiser: Prof. Andrew Glennerster (University of Reading)
Dr. Andrew Fitzgibbon (Microsoft Research, Cambridge)
Prof. Hanspeter Mallot (University of Tübingen)
Meeting date: 1 - 3 July 2015
Meeting location: Microsoft Research Cambridge
21 Station Road, Cambridge CB1 2FB, United Kingdom
Sponsor:
EPSRC