Reconstruction of as-built building models with topological and semantic object properties for building operation and maintenance
Final Report Abstract
The ReconTOP project had the goal to work on an automated pipeline to determine the type of usage of spaces starting from point clouds. The main focus in that pipeline is the topology which is to be reconstructed from polygons and polyhedrons describing the boundaries of spaces. It was planned to use existing algorithms for point cloud processing. Based on the topological reconstruction, features should be identified as the basis for the computation of the type of usage of spaces. The analysis of existing algorithms for calculating polygons and polyhedra from point clouds has shown that they cannot achieve the quality required to determine the topology. This is specifically true for the geometry of corners, which are not adequately captured. Extensions of wall geometries do not allow the geometric identification of wall planes in adjacent rooms. On the other hand, the basic approach of reconstructing the topology based on a spatial decomposition works. It can be applied to both, to building components and to spaces in the case of sufficient accuracy. Topological properties can be the input for space usage determinations. The research showed in addition that an automated pipeline can be established. However, the research was restricted to the two-dimensional space. Floor plans have been computed from point clouds. The research project shows at the end two pipelines: • One pipeline is able to determine the type of usage of rooms based on existing algorithms. This pipeline requires partly human interaction for specific parameters to achieve the results. The influence of certain parameters on the results cannot always be predicted, so different calculations must be performed with different parameters. • The other pipeline uses modeled room polygons as its input. The accuracy of this data allows an automated determination of all topological properties. Topological parameters are determined for individual rooms as well as for floors. They are together with geometric properties input values for algorithms to determine the type of usage. This pipeline does not require additional user input so that it works in an automated way.
Publications
-
Feature-based decomposition of architectural spaces: outline of a procedure and research challenges, ICCCBE 2022, Cape Town, South Africa
Suter, G.
-
SFS-A68: A dataset for the segmentation of space functions in apartment buildings
Ziaee, A., Suter, G. & Barada, M.
-
SFS-A68: a Dataset for the Segmentation of Space Functions in Apartment Buildings. Proceedings of the 29th EG-ICE International Workshop on Intelligent Computing in Engineering, 319-329. EG-ICE.
Abbas Ziaee, Amir & Suter, Georg
-
Where is the end of the wall, ICCCBE 2022, Cape Town, South Africa
Gabler, F. & Huhnt, W.
-
Extracting topological features from room polygons based on a 2-dimensional space partitioning approach, EG-ICE 2023, London, UK
Gabler, F., Ziaee, A., Huhnt, W. & Suter, G.
-
SAGC-A68: A space access graph dataset for the classification of spaces and space elements in apartment buildings
Ziaee, A., Suter, G. & Keiblinger, L.
-
SAGC-A68: a space access graph dataset for the classification of spaces and space elements in apartment buildings, EG-ICE 2023, London, UK
Ziaee, A. & Suter, G.
-
SFS-A68: A dataset for the segmentation of space functions in apartment buildings
Ziaee, A., Suter, G., Barada, M. & Keiblinger, L.
-
Evaluating Automated Floorplan Generation: Benchmark on Residential Buildings, ICCCBE 2024, Montreal, Canada
Elsafty, A. & Hartmann, T.
-
SFC-A68: A dataset for space function and space access element classification in entire floors of multi-unit apartment buildings
Ziaee, A., Suter, G. & Keiblinger, L.
-
Simplification of Polygons from Point Cloud Data for Automated Floorplan Generation, in Tagungsband Forum Bauinformatik 2024, Technische Universität Hamburg, 2024, pp. 268-275
Carlone, A., Gabler, F.
-
Benchmark for Topological Spatial Assesment of Indoor Residential Buildings, EC3, 2025, Porto, Portugal
Elsafty, A. & Hartmann, T.
-
Multiview Space Function Classification in Apartment Buildings Using Image Deep-Learning Semantic Segmentation. Journal of Computing in Civil Engineering, 39(5).
Ziaee, Amir & Suter, Georg
