Mobile Banking Transaction Authentication using Deep Learning

Abstract

The use of smartphones is growing in tandem with the rapid advancement of mobile technologies. The most critical key to future mobile banking and market activation is a safe financial transaction. This study offers a safe authentication solution to strengthen the security of mobile banking applications using Convolutional Neural Network model to provide an embedded facial recognition model in mobile bank application. The System works with two computer vision models working together, the Firebase Machine Learning vision model to perform the face detection and preprocessing of the image, and the MobileFaceNet model to process, classify and transform into a data structure ‘savable’ by a database (an array of numbers). Data were collected by taking four pictures of each user, recording key facial features such as the eyes, brows, nose, lips, and ears in real-time with a smartphone's front facing camera. The System provides a user-friendly interface that was developed using Java programming language and Dart programming language. After being tested in a variety of scenarios, the system achieved very encouraging authentication accuracy on real faces. It is a promising application that can reduce insecurity in banking sector.