Sciweavers

Share
CVPR
2011
IEEE

Mining Discriminative Co-occurrence Patterns for Visual Recognition

8 years 1 months ago
Mining Discriminative Co-occurrence Patterns for Visual Recognition
The co-occurrence pattern, a combination of binary or local features, is more discriminative than individual features and has shown its advantages in object, scene, and action recognition. We discuss two types of co-occurrence patterns that are complementary to each other, the conjunction (AND) and disjunction (OR) of binary features. The necessary condition of identifying discriminative cooccurrence patterns is firstly provided. Then we propose a novel data mining method to efficiently discover the optimal co-occurrence pattern with minimum empirical error, despite the noisy training dataset. This mining procedure of AND and OR patterns is readily integrated to boosting, which improves the generalization ability over the conventional boosting decision trees and boosting decision stumps. Our versatile experiments on object, scene, and action categorization validate the advantages of the discovered discriminative co-occurrence patterns.
Junsong Yuan, Ming Yang, Ying Wu
Added 30 Apr 2011
Updated 30 Apr 2011
Type Journal
Year 2011
Where CVPR
Authors Junsong Yuan, Ming Yang, Ying Wu
Comments (0)
books