Proceedings of the
9th International Symposium for Geotechnical Safety and Risk (ISGSR)
25 – 28 August 2025, Oslo, Norway
Editors: Zhongqiang Liu, Jian Dai and Kate Robinson
Soil Layer Classification from Cone Penetration Test Data: A CPT-as-Image Paradigm
1Department of Civil and Environmental Engineering, The Hong Kong University of Science and Technology, Hong Kong SAR, China.
2Department of Geotechnical Engineering, Tongji University, Shanghai, China.
ABSTRACT
Soil classification is a fundamental task in geotechnical engineering, providing essential information for various applications such as foundation design, slope stability analysis, and earthwork construction. Traditional soil classification methods, such as borehole logging, are often limited by high costs and the small number of boreholes. Cone Penetration Test (CPT) data offers a more cost-effective and widely available alternative for soil classification. This study proposes a YOLOv8-based deep learning model for classifying soil layer via aCPT-as-Image paradigm, treating CPT data as images. The performance of the model is evaluated using precision, recall, F1 score, and mean average precision (mAP). The established model chieved a mAP of 0.986. The results demonstrate that the proposed model achieves high accuracy in soil classification for different soil type (e.g., clay, silt, and sand).
Keywords: Soil classification, Deep learning, CPT data, YOLOv8 model.
