A Novel Approach to Detect Indian High Security Registration Plates

In recent times, with the growing number of vehicles, road safety has become a paramount issue. To ensure safe and easy passage of vehicles on the road, it is important that we take help of certain techniques that ensure the road safety regulations are properly followed. The Government of India has mandated the usage of High Security Registration Plate (HSRP) to ensure safety of vehicles, since these plates are tamper proof and cannot be counterfeited easily. In this paper, we have proposed a novel approach to detect the HSRP on a car. The most prominent feature detectable through images of an HSRP is the “IND” text inscribed on it. We have used three License Plate (LP) detection models: Warped Planar Object Detection Network (WPOD-Net), You Only Look Once (YOLO) version 3 (YOLOv3) and version 5 (YOLOv5) to detect the license plate on a car. Once a plate is detected, we apply EasyOCR to determine whether the plate has “IND” inscribed on it or not. A comparative analysis is done on these three models and our experimental results show that WPOD-Net with EasyOCR gives and accuracy of 93.01%, making the proposed approach viable for HSRP detection.