Automatic Road Inventory Using a Low-Cost Mobile Mapping System and Based on a Semantic Segmentation Deep Learning Model

Road maintenance is crucial for ensuring safety and government compliance, but manual measurement methods can be time-consuming hazardous. This work proposes an automated approach road inventory using a deep learning model 3D point cloud acquired by low-cost mobile mapping system. The includes the width, number of lanes, individual lane widths, superelevation, barrier height. results are compared with ground truth on 1.5 km subset road, showing overall intersection-over-union score 84% segmentation centimetric errors parameters. lanes correctly estimated in 81% cases. proposed method offers safer more to tasks extended complex objects rules digitalization. has potential pave way building digital models from as-built infrastructure systems, making process efficient accurate.