Smart Atm Safety and Security Monitoring System

Authors: Ojas Kulkarni, Parth Kulkarni, Yadnesh Kulkarni, Prathmesh Gobare, Suraj Kodale, Dr. N. P. Bhone

Abstract: ATMs have become a vital part of our world financial infrastructure but they are still a very high risk target of criminal acts such as burglary, illegal hacking, card skimming, loitering and damage to machines. Old style CCTV systems can only record video information without real-time notification or intelligence; this paper proposes an "Intelligent ATM Safety and Security Monitoring System" built on Python, OpenCV, Flask and SQLite. This system is a software only AI based monitoring solution for real time identification of ATM security threats. It is able to detect the presence of single or multiple people around ATMs, detects people who loiter around ATMs based on the use of time threshold algorithms and detects virtual boundary crossing of restricted areas around the machine. There is an emergency panic alert system as well. All event are time stamped and stored in the SQLite data base which can be accessed using a web administered panel through Flask web service. The system was tested to have a mean time to detect below 2s and did not required a GPGPU so is very cost effective and can be used at remote or inner city ATMs. System testing of 5 various types of simulatedATMthreats all showed satisfactory result in each detection modules.

DOI: http://doi.org/10.5281/zenodo.20422740