Intelligent Road Monitoring: Advanced Damage And Pothole Detection With Yolov8

Abstract

This project aims to improve road safety and 
infrastructure maintenance by leveraging deep learning 
techniques to detect and categorize common types of 
road damage, including potholes, longitudinal cracks, 
transverse cracks, and alligator cracks. Utilizing the 
YOLOv8 object detection model, trained on the 
Crowdsensing- based Road Damage Detection dataset, 
this project enables accurate and real-time detection of 
road damage from both images and video feeds. By 
employing the YOLOv8 model, known for its robust 
accuracy and low-latency performance, this project 
achieves efficient real-time damage identification on 
both static images and continuous video streams, 
addressing a significant need in automated road 
inspection systems. The integration of this technology 
can streamline the process of monitoring extensive road 
networks, providing data-driven insights for timely 
repairs and reducing human intervention. 
Keywords: Road Damage Detection, YOLOv8, Real
time Detection, Automated Road Inspection