Maintenance of roads is an important part of city infrastructure and development, whereby potholes play a major role in causing accidents, vehicle damage, and financial losses. Traditional pothole detection involves manual inspection that is time-consuming, inefficient, and manpower-intensive. This paper reviews available techniques for pothole detection and proposes an AI-based framework intended for real-time detection, severity estimation, and report generation. The suggested system uses modern computer vision, artificial intelligence, and intelligent transportation systems developments to detect potholes in real time, rate severity, and predict maintenance needs.
