Hierarchical object categorization with automatic feature selection

In this paper, we have introduced a hierarchical object categorization method with automatic feature selection. A hierarchy obtained by natural similarities and properties is learnt by automatically selected features at different levels. The categorization is a top-down process yielding multiple labels for a test object. We have tested out method and compared the experimental results with that of a nonhierarchical method. It is found that the hierarchical method improves recognition performance at the level of basic classes and reduces error at a higher level. This makes the proposed method plausible for different applications of computer vision including object categorization, semantic image retrieval, and automatic image annotation.