Deep Learning – Based Automated Pothole and Road Crack Detection System

Abstract

Road surface damage such as Potholes and cracks is one of the major causes of road accidents, vehicle damage, and traffic congestion. Manual road inspection is time-consuming, costly, and often inaccurate. This project presents a Deep Learning-Based Automated Pothole and Road Crack Detection System that uses the YOLOv3- Tiny object detection algorithm and OpenCV to automatically detect potholes and road cracks from images. The proposed system is developed using Python, Flask, OpenCV, and YOLOv3- Tiny, Providing a web-based interface where users can upload road images for instant analysis. The trained deep learning model identifies damaged road regions with bounding boxes and confidence scores, helping road maintenance authorities detect defects efficiently. The system offers faster, accurate, and cost-effective road inspection, reducing manual effort and improving transportation safety. KEYWORDS: Deep Learning, YOLOv3- Tiny, OpenCV, Flask, Road Crack Detection, Pothole Detection, Computer Vision, Image Processing, Object Detection, Python.