Development of an automated surface crack detection and BIM-integrated management system for concrete bridges

Bridge inspection work typically requires inspectors to capture hundreds to thousands of images, consuming substantial time for review. This research developed an “Automated Crack Image Cloud Detection System” and the “Auto Predictor” application, enabling automatic crack identification and deterioration image recognition through uploads. This platform integrates with the “Bridge BIM Cloud Management System”, connecting crack information with three-dimensional models. Engineers can create BIM models based on structural design drawings, while inspectors can photograph cracks and integrate relevant information. The study utilized deterioration images from long-term bridge inspections in Taiwan, covering various real-world environmental conditions. Through effective deterioration labeling strategies and comparing YOLOv4 and YOLOv7 algorithms with recommended parameters, an optimal model was obtained for system implementation. Research results demonstrate that the “Automated Crack Image Cloud Detection System” successfully identified cracks in bridge inspections and short beam shear test specimens. The deep integration with the “Bridge BIM Cloud Management System” enables automatic component crack identification and generates location charts, providing decision-makers with intuitive visual data. The YOLOv7-based model achieved a mean Average Precision (mAP) of 87.64%, significantly improving bridge inspection efficiency and demonstrating exceptional application potential.