Abstract:
We propose an unsupervised, probabilistic method for learning visual feature hierarchies. Starting from local, low-level features computed at interest point locations, the method combines these primitives into high-level abstractions. Our appearance-based learning method uses local statistical analysis between features and Expectation-Maximization (EM) to identify and code spatial correlations. Spatial correlation is asserted when two features tend to occur at the same relative position of each other. This learning scheme results in a graphical model that allows a probabilistic representation of a flexible visual feature hierarchy. For feature detection, evidence is propagated using Nonparametric Belief Propagation (NBP), a recent generalization of particle filtering. In experiments, the proposed approach demonstrates efficient learning and robust detection of object models in the presence of clutter and occlusion and under view point changes. 1.
Citations
|
792
|
A combined corner and edge detector
– Harris, Stephens
- 1988
|
|
492
|
Object recognition from local scale-invariant features
– Lowe
- 1999
|
|
351
|
Object class recognition by unsupervised scale-invariant learning
– Fergus, Perona, et al.
- 2003
|
|
311
|
Scale & affine invariant interest point detectors
– Mikolajczyk, Schmid
- 2004
|
|
155
|
Estimating the dimension of a model
– Schwartz
- 1978
|
|
152
|
Hierarchical models of object recognition in cortex. Nat Neurosci 2
– Riesenhuber, Poggio
- 1999
|
|
138
|
Learning a sparse representation for object detection
– Agarwal, Roth
- 2002
|
|
111
|
Pictorial structures for object recognition
– Felzenszwalb, Huttenlocher
|
|
86
|
A bayesian approach to unsupervised one-shot learning of object categories
– Fei-Fei, Fergus, et al.
- 2003
|
|
76
|
Nonparametric belief propagation
– Sudderth, Ihler, et al.
- 2003
|
|
50
|
Tracking loose-limbed people
– Sigal, Bhatia, et al.
- 2004
|
|
44
|
Pampas: Real-valued graphical models for computer vision
– Isard
- 2003
|
|
12
|
Extending pictorial structures for object recognition
– Kumar, Torr, et al.
- 2004
|
|
11
|
Columbia object image library
– Nene, Nayar, et al.
- 1996
|
|
10
|
A hierarchical axis of object processing stages in the human visual cortex. Cereb Cortex 11
– Lerner, Hendler, et al.
- 2001
|
|
10
|
Unsupervised learning of combination features for hierarchical recognition models
– Wersing, Körner
- 2002
|
|
9
|
Object recognition with many local features
– Helmer, Lowe
- 2004
|
|
6
|
Distinctive features should be learned
– Piater, Grupen
- 2000
|
|
6
|
Face detection based on generic local descriptors and spatial constraints
– Vogelhuber, Schmid
- 2000
|
|
5
|
Hierarchical image probability (HIP) models
– Spence, Parra
- 2000
|
|
2
|
Differences in the coding of spatial relations in face identification and basic-level object recognition
– Cooper, Wojan
- 2000
|
