Iot Based Atm Surveillance System Using Deep Learning

Abstract

ATM( Automated Teller Machine) security is very important to people nowadays. ATM transactions are very simple and easy, but the machines and the areas around them can be vulnerable to theft and need to be guarded. Using real-time video surveillance via hidden camera will make ATM transactions much safer. Robberies occur pretty much in our everyday lives. This system deals with the development of a video surveillance automation framework in ATM machines and detects any form of possible criminal activity and therefore brings out a solution to current problem with the existing systems. Deep learning techniques can be used to achieve impressive results in the detection of the activities. The proposed system makes effective use of an algorithm such as Yolov3 which includes methodologies such as detecting objects such as weapons and eventually ending up identifying the action required to prevent fraud activities. Our goal is to build a real-time system using Camera to catch the person entering the system. If some weapons are identified by the system it will take necessary actions. It will send alert message to the nearest police station and corresponding bank through GPS and GSM. This system also produce an alarm with beep sound once theft or robberies are recognised. The DC motor is used for closing doors of ATM.