Abstract :

A Crucial Part Of Biometric Authentication Systems Is Fingerprint Liveness Detection (FLD), Which Guards Against Presentation Assaults Utilising Fake Fingerprints Made Of Latex, Silicone, And Gelatin. Although Current Techniques Based On Multimodal Biometric Features Or Convolutional Neural Networks (CNNs) Show Promise, They Frequently Increase System Complexity, Computational Expense, Or Hardware Requirements. This Work Presents A Lightweight Deep Learning Framework For Reliable Fingerprint Liveness Detection In Order To Get Around These Restrictions. In Order To Extract Both Global Liveness Cues Like Perspiration Dynamics And Texture Irregularities As Well As Fine Ridge-level Features Like Pore Distribution And Distortions, The Proposed System Uses An Effective Object Detection Model With An Improved Backbone And Decoupled Detection Head.The System Only Uses Fingerprint Images, Which Ensures Hardware Simplicity While Maintaining High Discriminative Strength In Contrast To Multimodal Techniques That Need Additional Biometric Data. Using A Cosine-annealed Adam Optimiser And Sophisticated Regularisation, The Model Is Trained End-to-end On Benchmark Datasets To Enhance Generalisation And Minimise Overfitting. In Comparison To State-of-the-art Methods, Experimental Evaluations Verify That The Suggested Framework Delivers Superior Spoof Detection Accuracy, Robust Resilience To Novel Attack Materials, And Quick Inference Time. The Technology Provides A Workable And Scalable Way To Improve The Dependability Of Biometric Authentication In Realistic Situations Because To Its Lightweight Design And Versatility.

Published:

14-8-2026

Issue:

Vol. 26 No. 8 (2026)

Page Nos:

884-890

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite