Adaptive image classification based on folksonomy

In this paper, we present a novel adaptive image classification method for content-based image classification systems based on user defined tags and annotations. The proposed method utilizes low-level features and folksonomies for improved classification accuracy. Thus, users' perceptive semantics are modeled in terms of low-level features and they are combined with low-level image categorization adaptively. The proposed method has been thoroughly evaluated and selected results are illustrated in the paper. It is shown that, satisfactory improvements can be achieved with integrating folksonomies into classification scheme. Furthermore, it is a language-independent and low-complex method that can be used on various databases, languages and Content-Based Image Retrieval applications.