Smart Traffic Management Using Transfer Learning Approach for Improve Urban Mobility

Abstract

The increase in congestion on traffic lanes is a major problem hindering the development of an urban city. The reason for this is the increasing number of vehicles on roads leading to large time delays on traffic intersections. To overcome this problem and to make traffic control systems dynamic, several methods and techniques have been introduced throughout the years. The static traffic control systems worked on fixed timings which were allocated to each traffic lane and were not able to be altered. Also, there was no provision for counting and detection of pedestrians on the zebra crossings as well as the detection of emergency vehicles in traffic. We will explore several machine learning and deep learning models for the detection of vehicles and pedestrians in this review article, evaluate their viability in terms of cost, dependability, accuracy, and efficiency, and add some new features to improve the performance of the current system.