Iris Biometrics using Deep Convolutional Networks

Iris biometrics-based recognition and verification systems, is an important and well-studied problem. However, recent advances in deep convolution networks make them a viable tool for extracting meaningful attributes and the process of using such an algorithmically generated feature set to classify images may be applied to the iris recognition method. This paper explores the applicability of the Convolutional Neural Network to iris biometrics. It explores the benefits of automatically-generated features compared to the traditional method of hand-crafted features and uses algorithms based on fine tuned models of deep residual networks, to solve both the recognition and the verification problems, resulting in a 99.8% recognition rate.