Pothole detection using computer vision based techniques

A nation needs roads for transportation in order to provide opportunities for travel. Roads not only facilitate movement but also stimulate economic growth. Since potholes deteriorate automobiles over time and put passengers in danger, raising the chance of an accident, a variety of research models on pothole detecting systems may be found among various pavement distress. Research suggests that pothole detection may be automated with cost-effectiveness, accuracy, and speed by using computer vision techniques. Potholes are a major issue for transportation infrastructure. This research proposes a modified VGG16 network to detect potholes in real-time using YOLO-v4 and a faster RCNN with a different backbone. The development of a pothole detection system specifically designed to detect potholes was the aim of this study. It comprises of a technique for classifying potholes based on eyesight. When this device is switched on, a camera linked to it collects images from the outside environment and uses them to locate potholes. Similar to object recognition systems operating in real time, potholes are shown in real time and denoted by boxes. GPS coordinates of potholes are identified and added to a database in order to identify them. the web application-based solution that provides drivers with road conditions information prior to their journey. This system was built using the YOLO (You Only Look Once) algorithm to detect potholes in real time. In order to empower normal road users to make well-informed travel decisions, this initiative aims to reduce government barriers to data collection on subpar road infrastructure.