Research On The Application Of Living Face Recognition In ATM Machine

With the advancement of technology,artificial intelligence has gradually appeared in people's vision.The facial payment that can be seen everywhere in the streets has become the norm in life.While artificial intelligence has brought us convenience,it is also accompanied by security risks.In the news,we often see incidents of robbery withdrawing money from ATMs,and even some criminals use illegal means to steal customer account passwords and steal other people's property.Therefore,it is an urgent matter to set up a face detection system in the ATM scene to protect the safety of customers' property.The main research content of this paper is the application of live face recognition in ATM machines,and the main purpose is to protect the property safety of customers in ATM scenarios.This article mainly uses the following four modules to study the application of living body face recognition in ATM machines.They are face recognition module,nodding and shaking head module,blink detection module,and lip language module.The face detection module uses the improved YOLOv4 model for detection,which can recognize the faces that appear in the lens.When multiple faces are detected,it will prompt to prevent money from being robbed.Nodding and shaking head module uses improved camera projection algorithm(SNO)to detect nodding and shaking head.It can effectively put an end to fraudulent activities such as three-dimensional models,photos,videos and other illegal means,and can better realize face detection.The blink detection module uses the latest YOLOv5 network model and the improved gray-scale integral(G-EVA)model to complement two algorithms for blink detection.The lip language detection module uses the improved DFTN-50 model to perform lip language detection on customers.It can effectively prevent fraudulent methods such as three-dimensional models and videos,and further verify that it is the customer who is performing the withdrawal.After the research of the above four modules and the analysis of the experimental results,it is possible to successfully determine whether the target is a living body,which can well protect the safety of the customer's property.With the development of neural networks,facial payment has been seen everywhere in life.It is believed that in the near future,facial recognition and withdrawal will also be applied to ATM machines,and face live detection will better serve the people.