Detecting Credit Card Frauds Using Deep Learning and Face Detection

In this post-pandemic era, there has been a tremendous rise in online transactions. Along with this increase in frequency, there is an increase in fraudulent transactions. Additionally, a recent Nielsen Report estimates that over the next ten years, the card industry will lose 400 billion to fraud. Techniques of Machine learning are frequently used in classifying fraudulent credit card transactions because they operate under the notion that false patterns can be found by looking at previous transactions. But the process becomes difficult when it must take into consideration shifting customer behaviour and fraudsters' inventiveness in creating new fraud patterns. Systems for detecting fraud can use deep learning to uncover anomalies. With the ability to handle both big data and real-time data, deep learning is also superior to current machine learning techniques. We aim to provide a solution to solve this problem of credit card fraud detection by leveraging the technology of face detection and combining it with deep learning, thus proposing a two-step verification-based approach. Broadly, in the proposed approach, the user has to register using his card details and then capture his face. Then login to the system, enter the card details and the system will prompt for face capture once the face and the card details are captured they are verified with the already stored data in the database. If the verification passes then the transaction is successful otherwise the user is logged out of the system.