Key points are not available for this paper at this time.
Concrete bridges are vital infrastructure assets, yet their inspection often relies on labour-intensive, time-consuming, and sometimes subjective visual assessments.This study addresses these challenges by harnessing the power of Artificial Intelligence (AI) and Deep Learning (DL) for streamlined bridge inspection.Building upon the limitations of traditional methods, an enhanced YOLOv8s model is developed and trained on a refined CONBRID-YOLOv8 dataset.This dataset is specifically designed to minimize false positives, a common issue in concrete bridge defect detection.The integration of real-time data visualisation tools further empowers inspectors to optimize maintenance planning, ultimately enhancing bridge safety and longevity.The model exhibits exceptional performance in detecting and classifying prevalent concrete defects such as cracks, spalling, exposed bars, corrosion stains, and efflorescence.Through rigorous experimentation and analysis, the new model achieved a strong F-1 score of 0.75 and a mAP of 0.738 after 300 epochs.Real-world field testing underscores the model's practical effectiveness.Pioneering data visualisation techniques provide inspectors with the tools to rapidly interpret complex results and confidently prioritise maintenance strategies.This AI-powered approach represents a significant advancement in bridge inspection practices.By addressing the limitations of traditional methods & existing DL models, this study offers a more efficient, accurate, and objective solution for ensuring the safety and longevity of critical infrastructure assets.
Saseethar et al. (Mon,) studied this question.