CrackScan: Enabling Intelligent Edge Inspection with UAVs for Structural Health Monitoring

The assessment of structural well-being is critical in ensuring the safety, durability, and operational effectiveness of critical infrastructure. The importance of essential infrastructure maintenance and inspection has grown in response to the possible risks associated with structural deterioration. Manual inspection and measurement of cracks is time-consuming, labor-intensive, and sensitive to error. To address these issues, the use of Unmanned Aerial Vehicles (UAVs) in infrastructure maintenance appears to be an appropriate solution. The methodology involves using UAVs to capture high-resolution images of infrastructure surfaces, integrating real-time anonymization techniques, such as automated face blurring, annotating these images for crack locations and reference measurements, selecting and customizing a deep learning model YOLOv8, estimating crack dimensions through pixel-to-physical unit conversion, classifying the severity of the crack, and producing quantified results demonstrating the model’s accuracy of 98.29%. Engineers and specialists can simply deploy the proposed system to conduct structural inspections while quickly navigating complicated zones of infrastructure and allowing precise crack detection, producing the necessary findings for effective structural health assessment.