Road potholes remain a significant infrastructure and public safety concern, as conventional reporting systems rely heavily on manual verification and inspection processes that lead to delayed maintenance actions. This paper presents a comparative evaluation of the YOLOv11m and YOLOv11n object detection models for an AI-based pothole detection and reporting system. Both models were trained and evaluated under identical configuration settings using the publicly available Multi-Weather Pothole Detection (MWPD) dataset consisting of 3,087 images captured under diverse environmental conditions. Model performance was assessed using mean Average Precision at IoU 0.5 (mAP50), precision, recall, precision–recall curves, and confusion matrix analysis. Experimental results show that YOLOv11m achieved better performance metrics, recording mAP50 of 0.76692, precision of 0.76728, and recall of 0.7020, compared with 0.74843, 0.75973, and 0.6911 respectively for YOLOv11n. Although the performance differences were relatively small, YOLOv11m demonstrated a better overall balance between pothole localization accuracy and false detections reduction, making it the preferred model for integration into the proposed reporting platform. The developed prototype automatically analyses user-submitted road images, verifies pothole presence, displays detection results, and records reporting information. The results demonstrate the feasibility of integrating deep learning-based object detection into municipal road reporting workflows to improve the efficiency, consistency, and reliability of pothole verification.
| Item Type: | Article |
|---|---|
| Creators: | Creators Email / ID Num. Mohd Zaki, Aisyah Hannah UNSPECIFIED Ishak, Norlela UNSPECIFIED Sarnin, Suzi Seroja UNSPECIFIED Dohad, Mohd Rizal UNSPECIFIED |
| Subjects: | T Technology > TA Engineering. Civil engineering > Applied optics. Photonics > Optical data processing T Technology > TE Highway engineering. Roads and pavements > Pavements and paved roads |
| Divisions: | Universiti Teknologi MARA, Shah Alam > Faculty of Electrical Engineering |
| Journal or Publication Title: | Journal of Electrical and Electronic Systems Research (JEESR) |
| UiTM Journal Collections: | UiTM Journals > Journal of Electrical and Electronic Systems Research (JEESR) |
| ISSN: | e-ISSN : 3030-640X |
| Volume: | 29 |
| Number: | 1 |
| Page Range: | pp. 54-60 |
| Keywords: | Computer Vision, Object detection, Pothole detection, YOLOv11m, YOLOv11n |
| Date: | 31 October 2026 |
| URI: | https://ir.uitm.edu.my/id/eprint/149117 |
149117.pdf
