Deep Learning-Based Vehicle Detection and Classification in Traffic Management for Intelligent Transportation Systems

Abstract

For intelligent transportation systems (ITSs) and planning that makes use of exact location intelligence, accurate vehicle classification and detection are topics that are becoming more vital. Although computer vision and deep learning (DL) are smart techniques, there remain issues with effective real-time detection and categorization. The requirement for a large training dataset and the domain-shift problem are two prevalent issues in this area. This research proposes the use of the YOLOv3 (you only look once) algorithm to provide an effective and efficient framework for vehicle recognition and classification from traffic video surveillance data. Along with the other deep learning-based algorithms like faster RCNN and VGG16 pre-trained model, a machine learning model using bag of features (BoF) + support vector machine (SVM) is also compared and analyzed for detecting and classifying vehicles.