Fingerprint Liveness Detection Using Deep Learning

Biometric authentication is essential in today's world, and fingerprint recognition is widely used. However, fake fingerprints or spoofed fingerprints can be created using various materials, leading to authentication issues. Convolutional Neural Networks (CNNs) have shown promise in this field. Numerous research articles have put forward diverse approaches for detecting the liveness of fingerprints, in order to counteract the increasing use of fraudulent fingerprints for malevolent intentions. A Convolutional Neural Network (CNN) is a deep learning neural network used for image recognition and classification tasks, with building blocks including convolutional layers, pooling layers, and fully connected layers. The Adam optimizer is a widely used optimization algorithm in deep neural network training. It computes personalized adaptive learning rates for individual parameters by estimating the first and second moments of the gradients. Categorical cross-entropy is used for multi-class classification problems with more than two classes, while binary cross-entropy is used for binary classification problems with two classes. The CNN model used in the study achieved high training accuracy but lower validation accuracy. However, using the Adam optimizer and binary cross-entropy loss function with early stopping at 25 epochs led to improved performance with high training and validation accuracies and successful matching of fingerprint images.