ABSTRACT
Iris Biometrics recognition is one of the frequent identification techniques. It can provide a high degree of assurance to a person’s identity. In an iris recognition system feature extraction is the most important stage. Various types of feature extraction techniques are used to carry out Iris recognition systems. Many hand-made feature extraction techniques are involved in the production of biometric systems by experts. Thanks to the introduction of deep learning techniques in object recognition problems, CNN (Convolutional Neuron Network) has been able to acquire a lot of attention in the field of the iris recognition system. This paper proposed a productive iris recognition scheme with the transfer learning technique of InceptionV3. The implementation of the proposed system is based on the development of fine-tuning of a pre-trained CNN (Inception-v3) for the feature extraction and classification. The proposed methodology is being tested with a well-known IIT-D Iris dataset. The proposed methodology allows for an accurate recognition rate.
