Automatic Road Damage Detection System and Location Indicator to Government Portal for Immediate Maintenance
Damages on road surfaces like potholes and cracks influence roads safety, automobile performance, efficiency of transportation. The old fashioned road inspection systems are based on the use of manual survey, which is time consuming, expensive and not acceptable in continuous monitoring. In addition, most of the current automated systems report deterioration of roads without referring to the traffic density, therefore prioritizing maintenance inefficiently. This paper aims to resolve this drawback by introducing a proposal of an automatic road damage detection and prioritization system based on deep learning and cloud integration. YOLO object detector algorithm analyses the live camera video on the road and approximates the density of traffic, allowing high-priority roads to be identified. A Convolutional Neural Network (CNN) is then used to identify and categorize road damages i.e. potholes and cracks and determine their severity. The identified damage information and the position of the damage are sent to ThingSpeak cloud platform where they are real-time monitored and visualized with the help of the web-based dashboard. The experimental outcomes reveal that the given system can achieve proper detection, stability in real-time testing, and prioritization in maintenance issues in various environmental conditions. The system lowers the compliance cost through manually checking and it offers scalable smart road infrastructure management solution.