Genuinity Detection of People: A Comparative Analysis on HOG and One Shot Learning

Abstract

Financial companies that provide monetary services like loan approval, credit cards etc. need high assurance of the genuinity of customers which can be done by matching the information provided by the user and the details on their respective Identity cards. Manually checking up such details can be a tedious task and also lead to errors.This paper discusses a novel approach to check the genuinity by using different techniques of face recognition namely, Dlib frontal face, one shot learning and face matching techniques like cosine similarity. This paper also discusses the problems faced during the implementation of these techniques. The data set used, contains images of Indian Government Identity proofs like Aadhaar card, PAN card and Driving License. The basic idea is to crop the face from the images (ID cards) and match them.