Convolutional neural networks for 5G-enabled Intelligent Transportation System : A systematic review

Abstract

Modern 5G-enabled Intelligent Transportation System (ITS) provides comfort and safety to the end users by using various models and techniques most of which are based on the machine learning-based techniques. However, a large number of issues such as congestion control, safety and security, traffic management exist in modern ITS for which AI-based techniques can be used. From the existing literature, it has been observed that deep learning based techniques are widely used for object detection, localization, and classification in ITS which in turns provide unforgettable experience to both the users and drivers. Specifically, Convolutional Neural Networks (CNN) and their variants are widely used for object detection, localization, and classification in modern ITS. Motivated from these facts, in this paper, we have reviewed different types of CNN models used in modern ITS for traffic sign recognition, traffic light detection, vehicle classification and pedestrian detection. Various CNN based models and techniques are discussed and compared in comparison to the existing schemes using different performance evaluation metrics. From the in-depth survey provided in this paper, users can identify that which particular model can be used in modern ITS with respect to its merits over the others.