2008 · 209 citations · 25 references
ComputingComputer ScienceComputer VisionData MiningMachine LearningCategory TheoryClassificationRepresentation LearningSublinear AlgorithmsTaxonomic ClassificationTaxonomyCategorization ModelsCategory InductionCategory Tree StructureComplexityGreedy HeuristicsObject ClassificationVisual CategorizationVisual Concept Learning
The computational complexity of current visual categorization algorithms scales linearly at best with the number of categories. The goal of classifying simultaneously N <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">cat</sub> = 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sup> - 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sup> visual categories requires sub-linear classification costs. We explore algorithms for automatically building classification trees which have, in principle, logNcat complexity. We find that a greedy algorithm that recursively splits the set of categories into the two minimally confused subsets achieves 5-20 fold speedups at a small cost in classification performance. Our approach is independent of the specific classification algorithm used. A welcome by-product of our algorithm is a very reasonable taxonomy of the Caltech-256 dataset.
25
Robust Real-Time Face Detection
Paul Viola, Michael Jones · International Journal of Computer Vision · 2004 · 14.2K citations
Recognition-by-components: A theory of human image understanding.
Irving Biederman · Psychological Review · 1987 · 5.6K citations
A Comparison of Affine Region Detectors
Krystian Mikolajczyk, Tinne Tuytelaars, C. Schmid et al. · International Journal of Computer Vision · 2005 · 2.9K citations