Extraction and Processing of Geographic Data for the Automatic Generation of 3D Traffic Environments

The process of generating annotated data for deep neural networks is labor-intensive and time-consuming. To address this challenge, a potential solution lies in training the neural network within a simulated environment. Since creating large and detailed environments by hand is not straightforward and sometimes even unfeasible, the generation process is often automated. In this work, we propose a pipeline that enables the automatic generation of three-dimensional computer-generated worlds based on geographic data. To narrow down the scope of this vast domain, we concentrate the research on the development of traffic scenes. Therefore, the proposed pipeline combines data from the open-source platform OpenStreetMap and satellite imagery in the visual portion of the electromagnetic spectrum. Ultimately, a virtual traffic scene is successfully generated with a vast potential for various applications.