ISSUES IN 3D OBJECT RECOGNITION

1
ISSUES IN 3D OBJECT RECOGNITION Jean
Ponce Department of Computer Science and Beckman
Institute University of Illinois at
Urbana-Champaign
Joint work with Amit Sethi, David Renaudie and
David Kriegman
and Svetlana Lazebnik, Cordelia Schmid and
Martial Hebert
2
Human/Felix
Bug
Barbara Steele
Face
Joe
Camel
Problem
Recognizing instances
Recognizing categories
3
(No Transcript)
4
Variability
Camera position Illumination Internal parameters
Within-class variations
5
Question 1
Is it better to eliminate as many possible of the
parameters that govern appearance or is it better
to work with the raw pixels?
Note
We may know something about the shape and the
dimension of our image set.
This surface is not smooth.
6

• Brooks and Binford, 1981

• Sullivan and Ponce, 1998

• Invariants (Weiss, 1988 Rothwell et al., 1992
  etc.)

• Murase and Nayar, 1992

• Schmid and Mohr, 1996

7
Human
Bug
??
Face
Camel
8
Question 2
What is an appropriate object representation for
describing people, animals, chairs, boats, shoes,
etc. ??
or
Do we really believe that local pixel
signatures and their geometric/statistical
relationships are sufficient?
9
The Blum transform, 1967
Generalized cylinders Binford, 1971
10
Zhu and Yuille, 1996
11
Question 3
How do we construct object descriptions from
images? How do we segment images? How do we
compute our feature vectors?
12
Forsyth, 2000
13
Question 4
How can we formalize the object recognition
process?
or
What should the corresponding optimization process
try to optimize?
14
(No Transcript)
15
(No Transcript)
16
The dual
17
d3
The trace
d3
d2
d1
d2
d1
The pedal curve
18
d3
d2
d1
3
3
3
q
19
d3
Occluding contour
d1
Silhouette
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
Dim.
25
(No Transcript)
26
Question 5
How can we effectively deal with clutter ?
27
(No Transcript)
28
q
29
q
3
Frontier point
3D/4D
Baseline
(Cipolla, Åström and Giblin, 1995)
30
Question 6
How do we recognize objects at the category
level?