Face verification with fully dynamic size blocks based on landmark detection
Mohammad Ibrahim, Md. Iftekharul Alam Efat, Shah Mostafa Khaled, Mohammad Shoyaib · 2014
Face recognition and verification is still a challenging problem due to several issues such as pose, facial expression, occlusion, imaging conditions, rotation, size and orientation. This paper addresses the problem of recognizing human faces despite the presence in pose and size variation. To handle these problems, we mainly focus on block size definition. Instead of uniform block we thus propose Dynamic Size Blocks (DSB) that is based on facial landmarks. Extensive experimentations are performed on Labeled Faces in the Wild (LFW) dataset with restricted original images. Our proposed landmark based DSB definition has achieved an accuracy of 74.08% on all images and 82.26% on dataset images excluding extreme pose variations.