TrailSurfNET: Automated Hiking Trail Surface Classification Using CNNs and OpenStreetMap Annotations

Finlay, Mark and Chiagoziem Onwuegbuche, Faithful (2026) TrailSurfNET: Automated Hiking Trail Surface Classification Using CNNs and OpenStreetMap Annotations. In: Artificial Intelligence and Cognitive Science. AICS 2025. Communications in Computer and Information Science (2950). Springer, Cham, Dublin, Ireland, pp. 190-201. ISBN 978-303225808-3

Full text not available from this repository.

Abstract

Reliable trail surface information is important for safe navigation and accessibility, yet OpenStreetMap (OSM) coverage is often incomplete or inconsistent. This paper introduces TrailSurfNET, a framework that combines Sentinel-2 satellite imagery with OSM annotations to classify hiking trail surfaces. We develop a data pipeline that harmonizes noisy OSM surface tags into five classes (asphalt, paved, gravel, mud/dirt, grass) and generates a balanced dataset across the UK and Ireland. Several Convolutional Neural Networks (CNNs) are benchmarked, including VGG and ResNet variants trained from scratch and with BigEarthNet pre-training. Deeper ResNet models trained from scratch perform best. TrailSurfNET provides a first large-scale, reproducible benchmark for trail surface classification in Ireland and the UK and demonstrates the potential of deep learning to enhance OSM with scalable surface classifications.

Item Type:	Book Section
Uncontrolled Keywords:	Deep Learning; OpenStreetMap; Remote Sensing; Trail Surface Classification
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science > Computer Systems > Information Storage and Retrieval Systems > Information Storage and Retrieval Systems - Geography > Geographic information systems > Geospatial data T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science > Computer Systems > Information Storage and Retrieval Systems > Information Storage and Retrieval Systems - Geography > Geographic information systems > Geospatial data Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Staff Research and Publications
Depositing User:	Tamara Malone
Date Deposited:	16 Sep 2026 18:46
Last Modified:	16 Sep 2026 18:46
URI:	https://norma.ncirl.ie/id/eprint/9932

Actions (login required)

View Item