Using boosted cascade of class specific features for identification of fungal species

Fungus belongs to the group of eukaryotic organisms. Many of which are pathogens of humans and other animals. More than 35 million people in the world are suffering from fungal nail infections only [1]. Around 1.5 million fungal species are estimated to be present on earth. Out of which only 5% have been formally classified [2]. Out of these around 150 fungal species are classified in the category of human pathogens. According to the analysis of death records from U.S National Centre of Health Statistics (NCHS), fungal infections were the seventh most common cause of infectious disease related mortality in 1992 [3]. So there should be methods for correct identification of these pathogens for proper medication. In this paper we have worked on identification of fungal species by extracting class specific features from the set of images and then applying a learning procedure based on boosted decision stumps. Using this approach we have shown that recall of up to 90 % can be achieved.