Combination of Face detection, Face Recognition and Gender Recognition using Evidence Theory

This paper presents an information fusion strategy for digital photograph semi-automatic indexing according to people´s identities. The final goal is to help users to annotate their photos in the way of people´s identities. We combine the data obtained from face detector, face recognizer and from gender recognizer. Our fusion process is named -FUSINDEX-and, it has been integrated in a Web-based application of photos sharing and browsing. The algorithms of FUSINDEX are based on the evidence theory of Dempster-Shafer. This theory allows us to manage the imprecision and the uncertainty of the data obtained by image analysis tools and, this has been approved by our evaluation results.