AI Culvert Inspection Preserves Utah's Infrastructure

CHANGE

UDOT Central Maintenance commissioned a study through the department’s Research and Innovation Division. A team from the University of Utah was selected to automate the interpretation of culvert inspection videos using advanced computer vision and deep learning techniques. Read research report “Automated Interpretation of Culvert Inspection Videos Using AI and Computer Vision” (UT-25.20)

They developed a binary classification model to identify defective frames. The process employs multiclass image classification models to classify five major defect types, and an object detection model capable of localizing and classifying defects.

To bridge the gap between model output and practical deployment, a graphical user interface (GUI) was created for each model type, enabling UDOT staff to analyze inspection videos, review condition ratings, and generate detailed summary reports, even without technical expertise.

RESULT

When tested on 56 real-world videos, the object detection algorithms correctly assessed culvert conditions in 84% of cases. This process takes seconds rather than hours and rivals the accuracy of human inspectors.

The system offers a scalable, cost-effective, and objective approach to culvert inspection, reducing manual workload and increasing the consistency and accuracy of infrastructure condition assessments. This automated evaluation  helps prioritize maintenance based on condition ratings.

Ultimately, this system has the potential to streamline culvert management workflows, minimize human error, and lower operational costs while supporting timely, data-driven infrastructure decisions.

Next Steps: Full implementation at UDOT is ongoing. With the creation of the new Asset Management Group and administrators centrally and in all four Regions, this new method of rating culverts is positioned to reduce costs and improve culvert inspections statewide.

With further refinement and expansion, such tools hold strong promise for broader application across transportation infrastructure monitoring. The Western Transportation Research Consortium (WTRC) has selected this research to expand its development and deployment beyond Utah.