Multi-level discriminative dictionary learning towards hierarchical visual categorization

For the task of visual categorization, the learning model is expected to be endowed with discriminative visual fea-ture representation and flexibilities in processing many cat-egories. Many existing approaches are designed based on a flat category structure, or rely on a set of pre-computed visual features, hence may not be appreciated for dealing with large numbers of categories. In this paper, we propose a novel dictionary learning method by taking advantage of hierarchical category correlation. For each internode of the hierarchical category structure, a discriminative dictionary and a set of classification models are learnt for visual cate-gorization, and the dictionaries in different layers are learnt to exploit the discriminative visual properties of different granularity. Moreover, the dictionaries in lower levels also inherit the dictionary of ancestor nodes, so that categories in lower levels are described with multi-scale visual infor-mation using our dictionary learning approach. Experi-ments on ImageNet object data subset and SUN397 scene dataset demonstrate that our approach achieves promising performance on data with large numbers of classes com-pared with some state-of-the-art methods, and is more effi-cient in processing large numbers of categories. 1