Einsatz von Deep Learning zur automatischen Detektion und Klassifikation von Fahrbahnschäden aus mobilen LiDAR-Daten / Deep Learning for Automatic Detection and Classification ofRoad Damage from Mobile LiDAR Data

Abstract

In the context of automated data analysis, convolutional neural networks and the use of deep learning approaches have become state of the art. In the field of road condition assessment and evaluation, the performance of deep neural networks for the analysis of camera image data has already been demonstrated. For the first time, this methodology is to be applied to high-precision mobile LiDAR data of the Fraunhofer Pavement Profile Scanner in the form of 2.5D surface models in order to realize automatic road damage detection and classification on the basis of radiometric and geometric features. Thus, an automated detection of road damage in the form of precisely located geo objects is possible.