Traffic Sensing Using Data-driven Surveillance Cameras: A Survey of Methodologies, Technological Advances, and Deployment Challenges

抄録

Evidence-based decision-making in transportation planning is essential for sustainable development in developing countries, yet reliable traffic data remain scarce. Recent advances in data-driven computer vision offer improved traffic-sensing capabilities, but remain underexplored in resource-constrained contexts. This paper systematically reviews artificial intelligence (AI)-based traffic analyses from the perspectives of transportation planning and traffic engineering, highlighting hardware and methodology developments, applications, and challenges. It uncovers critical gaps in the current research landscape and proposes potential directions for future studies to adapt these solutions to local contexts better. This article seeks to guide transportation professionals in selecting suitable methods for deploying AI-driven traffic sensing technologies effectively.