Automated vehicle security system using convolutional neural networks and support vector machine

Abstract

With the rise in the infrastructure in the global economy, there is a need to impact the growth of security systems such as enhancing the security of vehicles at public places, societies or places with crowd. This could be done by keeping up with the monitoring of vehicles through vehicle License Plate Recognition (LPR). Since the numbers of vehicles are increasing on road day by day, it is essential to bring automation in its detection and recognition procedure. The objective of this presented work is to model a real time application to recognize license plate from a vehicle at parking of any society or public places via surveillance cameras. This paper mainly focuses on implementing the concept of component security which is marked by the presence of a blended system with car license plate recognizer as well as face recognizer recognizing its real owner. In proposed Automated Vehicle Security System (AVSS), the achievable model accuracy for Automated LPR model is 94% marked with the use of Tyserract for character recognition and model accuracy for facial recognition is raised to a mark of 83%. This model provides remarkable results and a need of another system where the owner or permitted drivers for a vehicle are mapped to vehicle license plate which could be made to use as collaboration to make it a real life deployable application.