The aim of the study was to evaluate the technical feasibility of fully automatic recording of relevant railway infrastructure during travel with the aid of a colour camera mounted on the front of the train. The focus was on the video real-time capable implementation of a combined detection and tracking algorithm and first tests based on recorded camera footage.
|Translated title of the contribution||Kamerabasierte Objekterkennung mit tiefen neuronalen Netwerken zur automatischen Erfassung der Gleisinfrastruktur|
|Journal||ETR Eisenbahntechnische Rundschau|
|Publication status||Published - Apr 2019|