A Review Paper on Retrieval Magnets for Facial Duplication by Search Based Face Annotation

Search Based Face Annotation (SBFA) is an effective technique to annotate the weakly labeled facial images that are freely available on World Wide Web. The main objective of Search based face annotation is to assign correct name labels to given query facial image. One challenging problem for search-based face annotation scheme is how to effectively perform annotation by exploiting the list of most similar facial images and their weak labels that are often noisy and incomplete. A large portion of photos shared by users on the Internet are human facial images. However some of these facial images are tagged with names properly, but many of them are not tagged properly. Also the duplicate facial images cannot be annotated. This problem has motivated to study a new technique called as auto face annotation which aims to annotate facial images automatically. In this paper, an effective unsupervised label refinement (URL) approach is proposed for refining the labels of web facial images using machine learning technique. The learning problem can be formulated as a convex optimization and develop effective optimization algorithms to solve the large-scale learning task efficiently. This paper also addresses the issues of duplicate human names and explores supervised/semi-supervised learning techniques to enhance the label quality. The results obtained from the proposed ULR algorithms will significantly boost the performance of the promising SBFA scheme.