Abstract :

In The Contemporary, Rapidly Changing Financial Ecosystem, Securing Digital Transactions Has Transitioned From A Technological Preference To An Absolute Security Imperative. Traditional Automatic Teller Machine (ATM) Ecosystems Rely Heavily On Physical Debit/credit Cards Embedded With Magnetic Stripes Or Electromagnetic EMV Microchips. These Physical Vectors Are Highly Susceptible To Malicious Exploitation, Including Hardware Skimming, Terminal Shimming, Shoulder Surfing, Card Cloning, And Physical Theft. Consequently, Commercial Banking Systems Incur Billions Of Dollars In Losses Annually Due To Card-based Financial Fraud. To Address These Critical Systemic Vulnerabilities, This Project Details The Conceptualization, Architectural Design, And Functional Implementation Of A Machine Learning Based Secure Cardless Financial Transaction System Using Multi-Factor Authentication (MFA). The Proposed System Completely Eliminates The Necessity For A Physical Card At The ATM Kiosk, Substituting It With A Robust Three-factor Authentication Pipeline, Reinforced By An Inline Machine Learning Fraud Classification Engine. The Authentication Sequence Begins With A Biometric Validation Stage Where The Clients Live Face Is Captured Via The ATM Kiosks Webcam. To Ensure High Reliability Under Variable Environmental Lighting Conditions Typical Of ATM Kiosks, The Captured Frame Is Processed Using A Multi-pass Viola-Jones Face Detection Algorithm Coupled With Contrast Limited Adaptive Histogram Equalization (CLAHE). Following Successful Face Cropping, A Deep Convolutional Neural Network (CNN) Using A Pre-trained ResNet-50 Model Extracts A Highly Discriminative 2048-dimensional Feature Embedding Vector. This Live Feature Vector Is Compared Against The Clients Enrolled Template Stored In The Database Using Cosine Similarity.

Published:

30-7-2026

Issue:

Vol. 26 No. 7 (2026)

Page Nos:

1262-1266

Section:

Articles

License:

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

How to Cite