Binary Gender Classification of African Fingerprints using CNN

Maruthukunnel Jacob, John (2023) Binary Gender Classification of African Fingerprints using CNN. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration manual) Download (3MB) | Preview

Abstract

Fingerprints, one of the most popular biometric authenticators, can distinguish between genders. The primary difficulty in fingerprint classification is time-effective models. In this paper, the author compares the binary gender classification performance of VGG-19, VGG-16, InceptionV3, and ResNet-50 for fingerprints. Traditional deep networks perform poorly and take longer to train because of their narrow depth range. In addition, they have to deal with model overfitting because there is a lack of fingerprint data. To solve the issues with insufficient fingerprint data, data augmentation techniques rotate, zoom, and flip is implemented. Transfer learning is used to pre-train the four CNNs to accelerate model training. Evaluation of the proposed model is done by testing, training accuracy, and loss. The VGG-19 model performed the best with a testing accuracy of 71.9% along with VGG-16 with 72.3% in comparison to InceptionV3 with 67.3%, and ResNet-50 with the lowest accuracy of 60.8%.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Muntean, Cristina Hava UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science T Technology > TK Electrical engineering. Electronics. Nuclear engineering > Biometric Identification
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Tamara Malone
Date Deposited:	19 May 2023 16:41
Last Modified:	19 May 2023 16:41
URI:	https://norma.ncirl.ie/id/eprint/6615

Actions (login required)

View Item