Real-Time Detection and Monitoring of Structural Cracks Using ConcreteCrack

Structural crack detection is a critical task in infrastructure monitoring and maintenance, as early identification of cracks can prevent severe structural damage and reduce maintenance costs. In this work, we propose ConcreteCrack , a YOLOv11n-based detection framework enhanced with hierarchical feature extraction (HGStem), multi-scale feature fusion (HGBlock), and dynamic feature alignment (DynamicAlignFusion) modules to accurately detect cracks of varying sizes, shapes, and orientations. Extensive experiments on benchmark crack datasets demonstrate that our method outperforms several state-of-the-art object detection algorithms, achieving high Precision, Recall, F1-score, and mAP, while maintaining real-time inference speed. Furthermore, Grad-CAM visualizations validate the interpretability of the model by highlighting actual crack regions, ensuring reliable detection even in complex scenarios. The proposed approach provides a robust and efficient solution for automated structural crack monitoring, enabling safer and more effective infrastructure inspection.