Anomalous Behavior Detection in ATM using Artificial Intelligence

Abstract

This paper proposes a new supervised algorithm for detecting abnormal events in confined areas such as ATM rooms and server rooms. The objective of this work is to establish a robust technical foundation that supports a secure and convenient social infrastructure, with abnormal behavior detection using image processing being one of the key technologies. Abnormal behavior detection involves creating a model based on normal behavior data and identifying any behavior that deviates from this model as abnormal. However, collecting comprehensive abnormal behavior data in advance can be challenging. Therefore, the ability to detect abnormalities using a model built solely on normal behavior data is highly valuable for practical implementation.  This proposed work presents examples of abnormal behavior detection using image processing techniques applied to ATM surveillance videos. Additionally, typical examples of abnormal behavior detection through motion image processing are demonstrated. Furthermore, our approach enhances system security by verifying the identity of the cardholder during ATM transactions. The combination of image processing algorithms and supervised learning enables effective identification of abnormal events, contributing to a more secure and reliable social infrastructure.