Computer Vision Technologies and Prevention of ATM Machine Theft in India: The Role of Real Time Alert Generation

Abstract

In Indian Banking sector, Automated Teller Machine (ATM) is one of the most popular channels as it is known for providing self-service, convenience, and ease of access to various banking services. ATM locations have become the prime targets to carry out criminal activities like ATM machine thefts, Card shimming, Card trapping etc by fraudsters as they are exposed round the clock with readily available cash and with less or no security. While the monetary loss owing to ATM frauds is significant, the full impact of fraud on a Bank can be devastating like losses to reputation, goodwill, and customer relations. Many contemporary academic research studies are dealt with ATM surveillance footages to generate real time alerts to arrest ATM machine thefts by analyzing the characteristics of a fraudsters like covering the face with facemask, wearing a helmet, possessing lethal / non-lethal weapons, sensors etc but they could not find a place to implement in banking sector because their implementation is leading to increased false positive alerts and thus creating a reputational damage to the Banks. The proposed research method will decipher the ATM surveillance footages with the help of Computer Vision Technologies, Machine Learning models and various Statistical/mathematical measures to generate the alerts about the commencement of ATM machine theft by analyzing the characteristics of ATM machine instead of analyzing the characteristics of the fraudsters.