Electrical Engineering
      and Computer Sciences

Electrical Engineering and Computer Sciences

COLLEGE OF ENGINEERING

UC Berkeley

Discovering Efficiency in Coarse-To-Fine Texture Classification

Jonathan Barron and Jitendra Malik

EECS Department
University of California, Berkeley
Technical Report No. UCB/EECS-2010-94
June 12, 2010

http://www.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-94.pdf

We introduce a model for joint texture classification and segmentation that learns not only *how* to classify accurately, but *when* to classify efficiently. This model, combined with a complementary efficient feature representation that we describe, allows us to move beyond naive sliding-window classification strategies into sub-linear coarse-to-fine classification of an entire image. Recognition is formulated as a scale-space traversal through the image in which we can ``stop short'' at coarse scales, dramatically increasing both the speed and the accuracy of classification. Unlike other models, ours is constructed such that the classification produced when stopping-short is exact (that is, equivalent to the classification produced when not stopping-short), because coarse-to-fine efficiency is directly incorporated into the model. Classification is demonstrated on partially- and fully-annotated datasets of satellite and medical imagery.


BibTeX citation:

@techreport{Barron:EECS-2010-94,
    Author = {Barron, Jonathan and Malik, Jitendra},
    Title = {Discovering Efficiency in Coarse-To-Fine Texture Classification},
    Institution = {EECS Department, University of California, Berkeley},
    Year = {2010},
    Month = {Jun},
    URL = {http://www.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-94.html},
    Number = {UCB/EECS-2010-94},
    Abstract = {We introduce a model for joint texture classification and segmentation that learns not only *how* to classify accurately, but *when* to classify efficiently. This model, combined with a complementary efficient feature representation that we describe, allows us to move beyond naive sliding-window classification strategies into sub-linear coarse-to-fine classification of an entire image. Recognition is formulated as a scale-space traversal through the image in which we can ``stop short'' at coarse scales, dramatically increasing both the speed and the accuracy of classification. Unlike other models, ours is constructed such that the classification produced when stopping-short is exact (that is, equivalent to the classification produced when not stopping-short), because coarse-to-fine efficiency is directly incorporated into the model. Classification is demonstrated on partially- and fully-annotated datasets of satellite and medical imagery.}
}

EndNote citation:

%0 Report
%A Barron, Jonathan
%A Malik, Jitendra
%T Discovering Efficiency in Coarse-To-Fine Texture Classification
%I EECS Department, University of California, Berkeley
%D 2010
%8 June 12
%@ UCB/EECS-2010-94
%U http://www.eecs.berkeley.edu/Pubs/TechRpts/2010/EECS-2010-94.html
%F Barron:EECS-2010-94