Automated rebar classification from point clouds

초록

Accurate and efficient rebar inspection remains a challenge in reinforced concrete construction due to reliance on manual, time-consuming, and error-prone methods. This paper investigates whether LiDAR-based point cloud data combined with machine learning can enable reliable automated rebar detection and classification in complex construction environments. A methodology integrating geometric feature extraction, forward feature selection, classifier optimization, and adaptive neighborhood radius determination is developed and evaluated on real-world datasets. The results show that the proposed approach achieves up to 92% accuracy and 89% F1score, with consistent performance across different datasets and strong generalization capability. These findings are particularly relevant for civil, structural, and construction engineers, QA/QC professionals, and BIM/VDC practitioners who are seeking efficient and scalable inspection solutions. The proposed framework provides a foundation for future research on multi-class classification, real-time implementation, and integration with digital twin systems.

키워드