An Explainable Deep Learning Model for Fingerprint Presentation Attack Detection

Automatic fingerprint recognition systems stand as the most extensively employed for person identification as compared with systems based on other biometric traits. Their usefulness in a variety of applications makes them vulnerable to presentation attacks which can be performed by presenting an artificial artifact of a genuine user’s fingerprint to the fingerprint based recognition systems. Hence, presentation attack detection becomes essential to ensure the security of fingerprint-based recognition systems. This paper proposes a novel method that incorporates the concept of explainability for the enhancement of the classification performance of the deep learning model. The proposed method consists of two building blocks including a heatmap generator and a classifier. The heatmap generator highlights the key features and generates a heatmap that helps the classifier to learn its parameters in a better way. The proposed method is validated using benchmark LivDet 2011, 2013, and 2015 databases. The comparative analysis demonstrates the superior performance of the proposed model in terms of classification accuracy when compared to state-of-the-art methodologies.