Template Matching for Kinship Verification in the Wild

Abstract

Kinship verification is a new research problem in the field of computer vision. It has gained attention from different research communities due to the fact that parent/child facial resemblance could be utilized to perform kinship verification. This paper proposes kinship verification based on template matching. We present two different methods for template matching on specific facial parts of parent/child image pair. First method selects each facial part in parent image and computes similarity score with corresponding facial part in child image. It analyzes and compares role of each facial part in parent/child image pair for kinship verification. Second method performs fusion of all facial parts and selects the one with maximum similarity score. Experiments are performed on baseline KinFaceW-I/II datasets to show effectiveness of facial parts in kinship verification.