Insight%202 :%20Representations%20for%20object%20recognition

1
Insight 2 Representations for object
recognition and
classification 3D structure from texture 3D
structure from shading Material properties KU
Leuven GA Orban U Utrecht JJ Koenderink U
Toulouse J Bullier U Liverpool P Giblin Inria
SophiaAntipolis O Faugeras TU Stockholm JO
Eklundh
2
HUMAN PERCEPTION
SINGLE CELL
MONKEY PERCEPTION
3
University of Liverpool UNC Chapel
HillPeter Giblin/Gareth Haslinger
James Damon
Classification of singularities of
shade/shadow/contour for Lambertian surfaces,
including unfolding by viewer movement.
Geometrical interpretation. Mostly complete for
fold shade cases. Others in progress. Production
of versatile programs to realise these, for use
by other partners. This includes the possibility
of conflicting cues from shading and contour.
A simple semi-fold singularity contour green
and shade boun-dary cyan
Unfolding of a semi-beaks singularity by moving
the view
Left consistent and rightinconsistent cues from
shade and shape
4
(No Transcript)
5
(No Transcript)
6
Mega Insight Bio- Computer Vision modules
7
Segmentation, scale, framework Reconstruction of
3D structure Optic flow and control subject
motion Control of effector motion Recognition of
actions Recognition and Classification of
objects Perceptual learning Attention and context
effects Priors and expectations.
8
Modular Interdisciplinary Goal Driven Tutoring
post docs Cutting edge close to the market Need
new products research Improve between humans
and human- computer communication Brain inspired
systems for aging and disabled.