Paper published in a book (Scientific congresses and symposiums)
Unsupervised learning of visual feature hierarchies
Scalzo, Fabien; Piater, Justus
2005In Machine Learning and Data Mining in Pattern Recognition
Peer reviewed
 

Files


Full Text
mldm.pdf
Publisher postprint (478.55 kB)
Request a copy

All documents in ORBi are protected by a user license.

Send to



Details



Keywords :
computer vision; machine learning; visual feature hierarchies
Abstract :
[en] 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 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 constitutes a probabilistic representation of a flexible visual feature hierarchy. For feature detection, evidence is propagated using Belief Propagation. Each message is represented by a Gaussian mixture where each component represents a possible location of the feature. 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.
Disciplines :
Computer science
Author, co-author :
Scalzo, Fabien;  Université de Liège - ULiège > Electrical Engineering and Computer Science > INTELSIG
Piater, Justus ;  Université de Liège - ULiège > Electrical Engineering and Computer Science > INTELSIG
Language :
English
Title :
Unsupervised learning of visual feature hierarchies
Publication date :
2005
Event name :
International Conference on Machine Learning and Data Mining
Event place :
Leipzig, Germany
Audience :
International
Main work title :
Machine Learning and Data Mining in Pattern Recognition
Publisher :
Springer-Verlag Berlin, Berlin, Germany
ISBN/EAN :
978-3-540-26923-6
Collection name :
Lecture Notes in Computer Science 3587
Pages :
243-252
Peer reviewed :
Peer reviewed
Available on ORBi :
since 29 July 2008

Statistics


Number of views
108 (7 by ULiège)
Number of downloads
1 (0 by ULiège)

OpenCitations
 
6

Bibliography


Similar publications



Contact ORBi