Criminal Vehicle Identification using Computer Vision and Deep Learning

Abstract

As registered vehicles grow, Sri Lanka needs more than just staff to manage and implement legal and safety regulations. The Sri Lankan Police Department or any other responsible party may benefit from this paper's deep learning method for recognizing license plates of vehicles in Sri Lanka to identify criminal vehicles. Since Sri Lankan has no proper automated approach to identifying criminal vehicles on the way, a study is carried out to implement a technological solution for this problem. This developed system provides a resilient and efficient deep learning method using a mix of two Convolutional Neural Network (CNN) architectures. The resulting product provides accurate identification of criminal vehicles under several challenges.