Automatic Certificate Verification using Computer Vision

Abstract

The issuance of fake degrees and certificates is one of the biggest concerns in society. This kind of employment through fake certificates will lead to a whole set of workforces who are not properly educated or trained. This can result in poor growth of the organization. The existing method of verifying the certificates involves sending an email of the certificate to the institute, the organization would respond with a reply mail containing all the details of the student along with the verification result. The drawbacks involved in this approach are delays in the verification process and this method becomes complex if the number of certificates to be verified is huge. Our approach to this problem involves detecting the labels in the image and extracting corresponding fields to be verified from the original image as multiple cropped images, the text in these cropped images is recognized using the EasyOCR module available in python. The institute database is searched for a match with the extracted details and the results after querying are sent back. This entire verification process is implemented as client-server architecture for easy access.