What type of study is this?

This is a Quantitative Study study.

September 19, 2025

Automated Road Damage Detection Using UAVs and Deep Learning: A Scalable Solution for Infrastructure Maintenance

Key Points

Detecting road defects achieved a precision of 0.51 with deep learning methods and UAV imagery, improving inspection efficiency.
Results indicate feasibility with a mean Average Precision of 0.42 on aerial imagery from a university campus, demonstrating effectiveness.
The method uses YOLOv7 deep learning model on publicly available datasets, streamlining pavement inspections over large areas.
This scalable solution offers proactive monitoring of infrastructure, highlighting its potential for modern smart city applications.

Abstract

This study presents an automated pavement surface inspection approach using UAV imagery and the YOLOv7 deep learning model. The aim is to detect common surface defects such as longitudinal cracks, reflective cracks, alligator cracks, and potholes with high accuracy and efficiency. The system was trained on publicly available datasets and tested with aerial images captured at Hacettepe University’s Beytepe Campus. Evaluation results demonstrated a precision of 0.51, recall of 0.45, and mean Average Precision (mAP@0.5) of 0.42. These findings confirm the feasibility of integrating UAV platforms and deep learning for road defect detection. Unlike traditional methods, which are time-consuming and labor-intensive, this approach enables faster, scalable, and cost-effective inspections. The proposed framework contributes to safer and more sustainable road infrastructure maintenance by facilitating proactive monitoring and reducing operational burdens. Its ability to analyze large areas in real time makes it particularly suitable for modern transportation networks and smart city applications.

Mark Helpful

Bookmark

Relay