ConDense: Multiple Additional Dense Layers with Fine-Grained Fully-Connected Layer Optimisation for Fingerprint Recognition

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Citation (Scopus)

Abstract

Fingerprint recognition is now a common, well known and generally accepted form of biometric authentication. The popularity of fingerprint recognition also makes it the focus of many studies which aim to constantly improve the technology in terms of factors such as accuracy and speed. This study sets out to create fingerprint recognition architectures which improve upon pre-trained architectures - named ConDense - that provide stronger if not comparable accuracy in comparison to related works on the authentication/identification task. Each of these ConDense architectures are tested against databases 1A, 2A, 3A provided by FVC 2006. The ConDense architectures presented in this study performed well across the varying image qualities in the given databases, with the lowest EERs achieved by this study’s architectures being 1.385% (DB1A), 0.041% (DB2A) and 0.871% (DB3A). In comparison to related works, the architectures presented in this study performed the best in terms of EER against DB1A, and DB3A. The lowest EER for DB2A reported by a related work was 0.00%.

Original languageEnglish
Title of host publicationPattern Recognition and Artificial Intelligence - 3rd International Conference, ICPRAI 2022, Proceedings
EditorsMounîm El Yacoubi, Eric Granger, Pong Chi Yuen, Umapada Pal, Nicole Vincent
PublisherSpringer Science and Business Media Deutschland GmbH
Pages15-27
Number of pages13
ISBN (Print)9783031092817
DOIs
Publication statusPublished - 2022
Event3rd International Conference on Pattern Recognition and Artificial Intelligence, ICPRAI 2022 - Paris, France
Duration: 1 Jun 20223 Jun 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13364 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference3rd International Conference on Pattern Recognition and Artificial Intelligence, ICPRAI 2022
Country/TerritoryFrance
CityParis
Period1/06/223/06/22

Keywords

  • Convolutional Neural Networks
  • Fingerprint recognition
  • Transfer learning

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'ConDense: Multiple Additional Dense Layers with Fine-Grained Fully-Connected Layer Optimisation for Fingerprint Recognition'. Together they form a unique fingerprint.

Cite this