Integrative AI and UAV-based visual recognition with metaheuristics for automated repair cost analysis of bridge structural deterioration

Aging bridges urgently need maintenance, as many exceed their lifespans. Traditional inspections are manual, time-consuming, costly, and error-prone. This has prompted a shift toward integrating advanced technologies to automate inspection processes and provide more efficient and accurate maintenance solutions. This paper introduces a multi-stage automated inspection system for bridge maintenance designed to classify bridge components and accurately assess the type and extent of deterioration. Unmanned aerial vehicles (UAVs) capture high-resolution images of bridge components, enabling comprehensive visual data collection without requiring manual access to challenging or hazardous areas. The inspection process employs the Vision Transformer (ViT) model for precise image classification, while You Only Look Once (YOLO) is used for instance segmentation. To further enhance the system's effectiveness, the Pilgrimage Walk Optimization (PWO)-Lite algorithm is applied to optimize the detection of deteriorated areas and estimate repair costs. This integration improves structural assessments, extends bridge longevity, and benefits bridge management agencies.