Rebar Classification and Detection in Concrete Elements using GPR data and Deep Learning
Volume Title: ICASGE2025
Paper ID : 1093-ICASGE-FULL (R1)
Authors
1Structural Engineering Department, Faculty of Engineering, Cairo University, Giza, Egypt
2Construction Engineering and Management (CEM)
3Integrated Engineering Design Management Program, Faculty of Engineering, Cairo University
Abstract
Structural health monitoring is a critical process that involves the continuous assessment of a structure's condition to ensure its safety and integrity. This is achieved through the utilization of non-destructive testing techniques, such as Ground Penetrating Radar (GPR), which provides valuable information about a building's internal structure without causing damage. This research proposes a framework that integrates artificial intelligence algorithms with Python scripting to generate images capable of accurately identifying and classifying rebars in GPR data. The framework comprises four stages: 1) Dataset Creation, 2) Dataset Processing, 3) AI Model Development, and 4) Model Evaluation. The training data includes experimental samples collected from designed specimens that simulate rebar locations within concrete. Various deep learning models are evaluated to determine the most effective approach. Among these, the YOLO v8 model outperforms both Faster R-CNN and YOLO v7. The model is trained and tested on real-world data collected from a building to validate its accuracy and capability to classify rebar diameters. Integrating the YOLO v8 model with GPR data demonstrates high accuracy and efficiency in detecting rebar locations within concrete structures.
Keywords
Subjects