Machine Learning Techniques for Kinship Verification: A Review

Kinship verification is an automatic determining process of people relationships, if two or more individuals are in kin relation or not. Since the verifying of a kinship is the most challenging problem for many applications, where become beneficial in many fields such as investigation cases of missing people through war and natural disasters, biometric security and more. The DNA test is the most common surgical method to detect the kin relations between individuals of families but it does not work with some applications scenarios that need to real time results due to the DNA takes hours or days to give a result. Thus, with the progress of years the kinship verification entered the computer vision world to determining the relationships using machine learning (ML) algorithms such as deep learning, transfer learning techniques and others. Each part of the human body may have a significant embedded information (features) that extracted and analyzed for verification or recognition and classification for that individual. This paper presents a comprehensive review of the kinship verification methods used, datasets, features extraction and what the accuracy achieved.