ATMs have become a cornerstone for the day-to-day life of millions, but they are vulnerable to attacks. The global ATM Industry Association has reported a 12% increase in ATM crimes during 2017. Attacks often involve ATM users, potentially injuring them, and thereby causing them trauma. Since ATMs are mostly situated outside buildings and are made to use at all the times of a day and night, securing them is a challenge – and therefore banks are turning to more intelligent solutions. An outdoor vulnerability According to ATM Industry Association (ATMIA), there is an estimation of about 3.5 million ATMs in the world serving those in need of cash 24/7, 365 days a year. This, coupled with the fact that they hold hard cash, makes them an attractive target. Criminals target these machines, or more precisely, the people using them in numerous ways. They distract customers at ATMs in an attempt to take their cards or cash, or to steal their PINs for later use. At cases, they also install false card readers that sneak customer’s card details for reuse. On the other side of the story, banks need to deal with customers’ disputes. In the situation of a very small percentage of ATM transactions, customers question whether the transaction has been carried out as requested – or even dispute at all in real time for any error while making the withdrawal. Thus, security precautions for ATMs are an important part of an overall security solution for financial institutions. When deep learning technology steps in Inside an ATM machine, two covert cameras are installed – one trained on the user and the other on the ATM panel. Deep learning technology embedded in the ATM security system can detect any ‘abnormalities’ in the facial scene in front of it, referring to the existing data patterns. So, if there is another face in the picture (for example someone looking over a user’s shoulder, or if the person wearing a mask), an alarm can be triggered in the security center. By using the same technology, the security system can also flag if the number pad is covered with a strip to steal PIN codes, or if a false card reader (or ‘skimmer’) is present to steal card details. All of these ‘smart’ alarms streamline the security monitoring processes, meaning that security personnel can react to real-time scenarios and would not waste time on false alarms. The footage can also provide evidence for any investigation. Hikvision solution Hikvision DeepinMind NVR – a robust deep learning ‘engine’ – takes information from the camera and analyses it using deep learning algorithms. This also ‘manages’ footage in conjunction with other NVRs and a video management system, providing a powerful toolset for security and business intelligence. Securing assets and mitigating risks through security solution is a lot easier with deep learning technology. Even ATMs outside a building can be safer, avoiding fraud and protecting customers every day.
