Abstract

The number of digital images is growing extremely rapidly, and so is the need for their classifification. But, as more images of pre-defifined categories become available, they also become more diverse and cover fifiner semantic differences. Ultimately, the categories themselves need to be divided into subcategories to account for that semantic re- fifinement. Image classifification in general has improved signifificantly over the last few years, but it still requires a massive amount of manually annotated data. Subdividing categories into subcategories multiples the number of labels, aggravating the annotation problem. Hence, we can expect the annotations to be refifined only for a subset of the already labeled data, and exploit coarser labeled data to improve classifification. In this work, we investigate how coarse category labels can be used to improve the classifification of subcategories. To this end, we adopt the framework of Random Forests and propose a regularized objective function that takes into account relations between categories and subcategories. Compared to approaches that disregard the extra coarse labeled data, we achieve a relative improvement in subcategory classifification accuracy of up to 22% in our large-scale image classifification experiments