A Smart Monitoring System for ATM Security and Surveillance

This paper presents an intelligent vision system modelled in Python using computer vision techniques like OpenCV, image processing techniques like Laplace operation and open-source frameworks like MediaPipe and MobileNet SSD to increase the level of security in ATM kiosks. Unmanned ATM kiosks are usually attached with a camera that records live footage which is checked manually and periodically, leading them to have late responses in emergency situations and making them the easier target for burglaries. Hence, early detection is of utmost importance to take preventive measures and safeguard the ATM and its users. As a consequence, it is important to deploy an automated and intelligent monitoring system to protect and secure the ATM machine from threats and inform the authorities at the earliest. This paper will demonstrate a system which allows only one person to use the ATM at a time. It will detect if the person using the ATM kiosk is displaying anomalous behaviour like moving out of the camera frame or scanning the ATM environment instead of using it and detect the presence of any harmful objects like hammer or rod. This proposed system will take anomalous situations of ATM kiosks into account and counter them by sending a signal to the warning system