YOLO-Based Framework for Detecting Suspicious Activities in ATMs

Abstract

The YOLO (You Only Look Once) algorithm is the basis of this research's novel approach to ATM security. After that, we suggest a deep learning model for real-time suspicious activity identification utilizing LO (Look Once) techniques. The proposed framework allows our system to detect and monitor security risks in an ATM environment using YOLOv5's sophisticated object detection. It has a detection accuracy of 94.3% and analyzes video in real-time for surveillance purposes. This suspicious activity includes cash trapping, card skimming, and strange human movements. Forty ATMs were used to showcase the technology. Executing at a rate of 30 frames per second with an average accuracy of 0.91. These findings show that conventional monitoring has made great strides forward. Better technique sensitivity is also provided by high detection, which has a false positive rate of just 0.4%.