Advancing Infrastructure Inspection with AI and 3D Data Platform Integration for Improved Damage Assessment and Enhanced Modeling

In this study, we introduce a three-dimensional infrastructure data platform developed by our team. The research underscores the fact that three-dimensional models, in contrast to two-dimensional ones, contain more information and enable a variety of analytical methods. The focus of this paper is on the various applications derived from the data of the three-dimensional infrastructure data platform. These applications include image processing AI for damage detection, generation of textual explanations, and the automated creation of Finite Element Method (FEM) models. The image processing AI first identifies the positions of cracks and corrosion damage in images, and then uses this information to form a three-dimensional plotted model. Additionally, an AI using image captioning technology has been developed to aid junior technicians and infrastructure managers who may lack expertise in damage assessment. This AI generates explanatory texts about the detected damage. Furthermore, this feature has been integrated into a web system, enabling access to the AI via mobile devices such as tablets at infrastructure inspection sites. The system improves its intelligence the more it is used and continually enhances its accuracy. Regarding the automated generation of FEM models, the AI constructs these models with the necessary precision using dimensions obtained from drawings or point cloud data. This enables a more comprehensive diagnosis of infrastructure, including mechanical behavior, which is often overlooked in traditional inspections and diagnoses. The paper also discusses how these AI models are centrally managed on a three-dimensional visualization platform and presents the results of this integration. The study demonstrates how leveraging AI and machine learning enhances accuracy and efficiency, significantly advancing the field of infrastructure inspection and diagnostics.