A MOBILE APPLICATION FOR ENHANCED ATM SECURITY AND EFFICIENCY: COMBINING HELMET DETECTION, FRAUD PREVENTION, AND REAL-TIME CASH MANAGEMENT
D.S. Kuruppu1*, K. Galappaththi2, G.M.C. Prabhashwara3, M.G. Nayanajith4, W.L. Gimhan5, and N.S. Madanayaka6
2,3,4,5,6Institute of Technology University of Moratuwa, Sri Lanka, 1Sri Lanka Institute of Information Technology, Sri Lanka
Session: Technical Session D
Abstract
Automated Teller Machines (ATMs) are essential for financial transactions but often lack robust security measures, especially in developing nations. In Sri Lanka, it is illegal to access ATMs while wearing helmets as it hinders identification in case of robbery or fraud. Additionally, customers are required to notify their banks if an ATM runs out of cash before the scheduled refill. To address these two issues a mobile application was developed by leveraging the Internet of Things (IoT) and Machine Learning (ML). A suite of technologies including Python, OpenCV, PyTorch, Firebase, and Flutter were incorporated to develop the application. The mobile application employs image processing and ML to detect individuals wearing helmets, triggering alerts for both customers and bank officials. OpenCV, a Python library for computer vision, is used for image processing and face detection to identify unauthorized wearers and alert bank officials in real-time. This feature enhances security by discouraging fraudulent activities and ensuring proper identification of users from fraudsters. The ML model trained for helmet detection reached a higher accuracy level of 99.94%. For efficient cash level monitoring, an IoT system integrates Infrared sensors with Arduino and is programmed with C++ to monitor ATM cash levels, providing proactive notifications to bank authorities before cash runs out, and minimizing service disruptions. The research contributes to improving ATM security and operational efficiency by combining advanced technologies with a user-centric design.
Keywords: ATM security, efficient ATM, IoT, Machine Learning, real-time cash monitoring
DOI: 10.64752/OFYT5742