Automatic creation of digital building twins with rich semantics from dense RGB point clouds through semantic segmentation and model fitting

M. Mehranfar, A. Braun, A. Borrmann

Research output: Contribution to conferencePaperpeer-review

3 Scopus citations

Abstract

Digital twins have emerged as a crucial tool for the operation and maintenance of buildings and infrastructure. A digital twin (DT) is a virtual replica of assets that provides valuable insights into real-time simulation, and monitoring. In this regard, laser scanners have become a critical component in the creation of DTs for built environments, owing to their ability to capture highly accurate point clouds of scenes. This paper presents an automatic algorithm for the creation of digital building twins with rich semantics and coherent geometry from the dense RGB point cloud. The proposed method aligns the capabilities of artificial intelligence (AI) methods in scene understanding with domain knowledge to overcome the data challenges. The results demonstrate the effectiveness of the proposed method to generate building DT models automatically with a mean error below 6 cm between the model's parameters reported by the facilities department and the parameters of generated models.

Original languageEnglish
StatePublished - 2023
Event30th International Conference on Intelligent Computing in Engineering 2023, EG-ICE 2023 - London, United Kingdom
Duration: 4 Jul 20237 Jul 2023

Conference

Conference30th International Conference on Intelligent Computing in Engineering 2023, EG-ICE 2023
Country/TerritoryUnited Kingdom
CityLondon
Period4/07/237/07/23

Fingerprint

Dive into the research topics of 'Automatic creation of digital building twins with rich semantics from dense RGB point clouds through semantic segmentation and model fitting'. Together they form a unique fingerprint.

Cite this