CC BY 4.0Zhou, YimingYimingZhouOsman, AhmadAhmadOsmanWillms, MarcMarcWillmsKunz, AlbrechtAlbrechtKunzPhilipp, SelinaSelinaPhilippBlatt, JanineJanineBlattEul, SimonSimonEul2023-11-172023-11-172023https://publica.fraunhofer.de/handle/publica/457001https://doi.org/10.24406/publica-217910.24406/publica-2179We present a conceptually simple and effective algorithm to detect user specified wireframes in a given indoor room image. Previous deep learning-based methods can produce great line detection results, however, they contain many redundant information for some cases. Hence, our method integrates semantic segmentation algorithm to control which part of the imagine should be detected. Segmentation is the task of clustering parts of an image together which belong to the same object class. According to the class information the proposed algorithm can show the desired results e.g. wireframe between walls and ceilings. Our method can give texture information and prepare for the following reconstruction.enWireframeDetectionsemanticsegmentationalgorithmDDC::600 Technik, Medizin, angewandte Wissenschaftenconference paper