On the use of supervised features for unsupervised image categorization: An evaluation

• We compared high- and low-level features for unsupervised image categorization.
• We verified that high-level features significantly outperform low-level features.
• We assessed how much the performance depends on the dimensionality of the feature vectors.
• We verified that a simple clustering on supervised features outperform strategies specifically designed for this task.

Recently, new high-level features have been proposed to describe the semantic content of images. These features, that we call supervised, are obtained by exploiting the information provided by an additional set of labeled images. Supervised features were successfully used in the context of image classification and retrieval, where they showed excellent results. In this paper, we will demonstrate that they can be effectively used also for unsupervised image categorization, that is, for grouping semantically similar images. We have experimented different state-of-the-art clustering algorithms on various standard data sets commonly used for supervised image classification evaluations. We have compared the results obtained by using four supervised features (namely, classemes, prosemantic features, object bank, and a feature obtained from a Canonical Correlation Analysis) against those obtained by using low-level features. The results show that supervised features exhibit a remarkable expressiveness which allows to effectively group images into the categories defined by the data sets’ authors.