Drone Photogrammetry and AI for Monitoring Paved Roads Distresses in Cameroon — A Case Study of Elig-Effa-Melen-EMIA Road Section in Yaoundé

Abstract

This study presents an innovative approach for automated pavement degradation monitoring in Cameroon, specifically focusing on the city of Yaoundé. We integrate drone photogrammetry for high-resolution images with artificial intelligence, leveraging a YOLOv11 model for precise detection and segmentation of various road surface degradations. A comprehensive Python workflow was developed to automate the entire processing chain, from orthomosaic tiling and AI model inference to georeferencing predictions and generating GIS-ready layers. The methodology was applied to a 2.5 km road section (Elig-Effa–Melen– Carrefour EMIA) in Yaoundé. The trained model achieved a F1-score of 77% with a precision of 86% and recall of 70% for the detection and F1-score of 81% with a precision of 86% and recall of 78% for the segmentation. The automated workflow processed the road section in 5 minutes 47 seconds, detecting 438 instances of degradation with a total affected surface area of 662.96 m², demonstrating significant time savings compared to days or weeks required for manual inspection. The generated degradation maps and an interactive dashboard provide objective, actionable insights for road managers, enabling proactive maintenance planning and optimized resource allocation. While acknowledging limitations such as the need for further validation with field surveys and the current focus on 2D surface data, this research highlights the scalability and cost-effectiveness of the proposed solution. It offers a promising pathway toward intelligent and efficient road network management in developing countries, by providing timely and accurate data to support data-driven maintenance strategies.

Keywords: Photogrammetry, Drone, Artificial Intelligence, Paved Roads, Degradations, YOLOv11.

Received Date: August 21, 2025              

Accepted Date: September 13, 2025            

Published Date: October 01, 2025

Available Online at https://www.ijsrisjournal.com/index.php/ojsfiles/article/view/458