Gender Identification Through Fingerprints with Deep Convolutional Neural Network
Buse Melis Özyıldırım, Serkan Kartal · 2016
Biometric features such as fingerprint, iris, face, etc. are mainly used in security systems. Fingerprints have been accepted as evidence in the court of law. Fingerprints provide key information about a person from gender to ethnicity. Gender identification is also important to decrease the time required for searching population of suspects. In this work, a novel method for identification of a gender from fingerprints is proposed. Unlike existing methods, proposed method utilizes deep learning. It consists of multi-convolutional layers, multi-pooling layers and two-layered MLP. It achieves 82% classification accuracy on NIST Special Database 4.