Search Based Face Annotation Using PCA and Unsupervised Label Refinement Algorithms

Abstract — Face recognition/detection presents a challenging problem in the field of image analysis and computer applications, and it is becoming more popular day by day because of its applications in various fields. Face annotation provide ways for recognizing facial images. Face annotation process is part of face detection, face recognition. Now a day’s research interest is in mining weakly-labeled facial images for resolving research challenges in computer vision. Search based face annotation framework is proposed to tackle the problems related to face image and label quality. This framework proposed Unsupervised Label Refinement algorithm to refine weakly labeled facial images. Clustering based Approximation (CBA) algorithm to improve efficiency and scalability. Experiment results show that the ULR algorithm boost the performance of the proposed search based face annotation framework