Kinship verification between two people by photos

Abstract

This paper examines the feasibility of using convolutional neural networks to determine kinship between two individuals based on their photographs.

In this work, the kinship verification problem is studied as a part of the image classification problem. Methods for solving the image classification prob-lem are explored and compared. During the research, models based on the ResNet50 and AlexNet architectures were developed. The performance of these two models is compared on a test set to determine the best model. The best model will be used for the implementation of the proof-of-concept application.