Search Based Face Annotation Using Weakly Labeled Facial Images

Automated face annotation aims to automatically detect human faces from a photo and further name the faces with the corresponding human names. Most of the images are obtained from World Wide Web (WWW), so there is possibility of getting noisy and incomplete images. This study tackle problem by investigating a search-based face annotation (SBFA) paradigm for mining large amounts of facial images. Given a query facial image for annotation, the idea of SBFA is to first search for top-n similar facial images from facial image database and then exploits these top-ranked similar facial images and their weak labels for naming the query facial image. Major challenging task of search based annotation scheme is how to perform effective annotation by exploiting the list of most similar facial images and their weak labels. Unsupervised label refinement (ULR) approach is used for refining the labels of facial images. In this proposed work unsupervised label refinement approach effectively solves the problem of weak labelling. A clustering-based approximation algorithm is proposed to improve the scalability of naming and images. In future it will evaluate a larger database to develop more efficient solutions. Keywords— Search based face annotation, unsupervised label refinement, clustering based approximation, weak label, auto annotation, face annotation.