Incorporating geometry information with weak classifiers for improved generic visual categorization
Gabriela Csurka, Jutta Willamowski, Chris Dance, Florent Perronnin
In this paper we introduce geometric information to improve the performance of a generic visual categorizer based on the bag of keypatches approach. More precisely we define a set of simple geometric relationships between interest points. Each relationship corresponds to a weak classifier and they are combined within a boosting framework. Experimental results are shown on a new challenging 10 class dataset.
ICIAP, Cagliari, Italy, 6-8 Sept 2005.