Real-Time Pothole Detection and Audio Warning System for Drivers

Potholes in roads creates a dangerous situation where vehicles get damaged and accidents happen. Manual detection methods are traditional, ineffective, laborious, and inefficient, requiring automation. In this paper we introduce a real-time pothole detection system and audio warning for drivers using YOLOv8, an advanced state of the art deep learning model designed for object detection. This system takes road pictures using a camera, processes them using the trained YOLOv8 model, and detects potholes in real time. Once detected, it produces only an audio alert in case of collision, so as to alert the drivers for road safety. The methodology suggests dataset collection, model training, extension of a Graphical User Interface (GUI) to manipulate the visualization graphically in real-time. We obtain high accuracy according to performance evaluation metrics like Mean Average Precision (mAP) and Intersection over Union (IoU). This system provides an affordable and scalable solution for real-time pothole detection, contributing to safer transportation infrastructure. Key Words: pothole detection, YOLOv8, deep learning, real-time detection, road safety, audio warning system