V3I7P87

Intelligent Concrete Crack Detection for Structural Health Monitoring

Rutuja Vijaykumar Taur^1*,  S. P. Sharma²  , L. S. Mahajan³

Abstract

This study investigates the efficient structural health monitoring, concrete fractures must be accurately and promptly detected since they are a significant signal of structural degradation. In this thesis, methods for automatic crack identification and classification in concrete structures using deep learning and image processing are presented. To improve edge identification and crack segmentation in noisy environments, a hybrid approach that combines the GrabCut algorithm with an enhanced Sobel filter is suggested. A graphical user interface for real-time crack assessment is created using MATLAB. For precise crack identification and classification, a modified Dual-Attention Faster R-CNN linked with ResNet-101 is also suggested. The efficacy and resilience of the suggested strategies are demonstrated by experimental assessment on both public and self-prepared datasets.

Keywords:

Concrete, Compressive strength, Strength, RCA