Marking Celebrity Faces Utilizing Annotation by Mining Weakly Labeled Facial Images

Abstract

Face annotation is a note or description added to the image for better understanding. It can help to improve better search due to detailed description. Face annotation in images and video enjoys many potential applications in multimedia information retrieval. Face annotation usually requires many training data labeled by hand in order to build effective classifiers. This is particularly challenging when annotating faces on large-scale collections of media data, in which huge labeling efforts would be very expensive. The goal of the system is to annotate the celebrity face images from childhood to recent face image and also to annotate the unseen faces in video. In particular, given a facial image, we first retrieve top n similar instances from facial image database. Initially the database contains images and description mapping of that image. Later the celebrity face image and/or the video that need to be processed will be considered. The detected face image will be processed with existing database. The matching result will produce mating annotation or null. For the video; that video will be converted to frames. This frame will act as an image. Further training can be same as for the celebrity face image. A know framework search based face annotation (SBFA) is there to mine weakly labeled facial images is used. Use of annotation will help for user to search desire image and video e.g. news videos. Automatic tagging of people will improve the search results. Also if system gets implemented in social network then it will overcome the drawback of current existing system which tags manually like Facebook.