Real-Time Crack Detection and Severity Assessment in Civil Infrastructure Using Deep Convolutional Neural Networks

Abstract

The structural safety and long-term durability of civil infrastructure systems such as bridges, tunnels, pavements, and buildings are significantly affected by crack formation and progressive structural deterioration. Conventional crack inspection methods primarily depend on manual assessment techniques, which are time-consuming, labor-intensive, and often limited in accuracy for large-scale infrastructure monitoring. This research presents a real-time crack detection and severity assessment framework for civil infrastructure using deep convolutional neural networks to improve structural health monitoring and automated damage evaluation. The proposed framework integrates deep learning–based image processing techniques with high-resolution visual data acquisition systems for accurate crack identification and classification under varying environmental and lighting conditions. Deep convolutional neural network (DCNN) models are employed to automatically extract crack features, detect structural defects, and estimate crack severity levels based on crack width, length, propagation pattern, and surface damage characteristics. Image preprocessing, feature enhancement, and data augmentation techniques are incorporated to improve detection accuracy and model robustness. The framework further supports real-time infrastructure monitoring through automated analysis and intelligent decision-making mechanisms. Performance evaluation is conducted using parameters such as detection accuracy, classification precision, processing speed, false detection rate, severity estimation reliability, and computational efficiency. Experimental analysis demonstrates that the proposed DCNN-based framework significantly improves crack detection performance and reduces inspection time compared with conventional image processing and manual inspection approaches. The findings further reveal enhanced capability in identifying fine cracks and assessing structural damage severity under complex operating conditions.