Paper Title
Caucasian and Asian Ethnicity Classification Using Iris as Biometric Modality and Deep Learning

Abstract
The rough texture of iris continues to be one of the strongest and scarcely researched soft biometric technique for AI based classification of ethnicity. Most research on ethnicity is done on eye images as compared to iris. As per our research till now this is the first work on ethnicity using iris on NDIRIS – 0405 dataset on reduced quality normalized data. Iris has a lot more information than the popular fingerprint or face. Being a unique specimen of its kind it is consistent throughout life. Iris patterns once formed during fetal development would persist and exhibit the recessive features in successive generations. It is established that certain similarities in the iris pattern between two individuals of the same ethnicity. Ethnic self-identification and belonging to a designated ethnic group are significant because they have been connected to majority of facets of human existence. The ethnicity classification on iris scans is the main objective of this study. We have handcrafted highly rich feature normalized images using Daugman’s approach and our feature extraction technique. As we worked on quality dataset with 3489 iris images our less complex deep learning Convolution Neural Network (CNN) model demonstrated highly accurate ethnic difference on the joint dataset of White and Asian ethnicities. Vast and very less used ND-Iris-0405 Data Set is utilized to evaluate proposed and presented model. In comparison to current methodologies and techniques, the effectiveness of the proposed CNN is demonstrated by the outcomes. In the rest paper we referred Caucasian as White as images in the database are labelled as white. Keywords - Biometrics Soft Traits, Iris, Ethnicity, Pattern Recognition, Deep Learning