Abstract
Abstract. 3D building modeling is a diverse field of research with a multitude of challenges, where data integration is an inherent component. The intensively growing market of BIM-related consumer applications requires methods and algorithms that enable efficient updates of existing 3D models without the need for cost-intensive data capturing and repetitive reconstruction processes. We propose a novel approach for semantic enrichment of existing indoor models by window objects, based on amateur camera RGB images with unknown exterior orientation parameters. The core idea of the approach is the parallel estimation of image camera poses with semantic recognition of target objects and their automatic mapping onto a 3D vector model. The presented solution goes beyond pure texture matching and links deep learning detection techniques with camera pose estimation and 3D reconstruction. To evaluate the performance of our procedure, we compare the estimated camera parameters with reference data, obtaining median values of 13.8 cm for the camera position and 1.1° for its orientation. Furthermore, a quality of 3D mapping is assessed based on the comparison to the reference 3D point cloud. All the windows presented in the data source were detected successfully, with a mean distance between both point sets equal to 3.6 cm. The experimental results prove that the presented approach achieves accurate integration of objects extracted from single images with an input 3D model, allowing for an effective increase of its semantic coverage.
Highlights
Building Information Modeling (BIM) is a diverse and multidisciplinary research subject with steadily increasing interest and demand (Czerniawski and Leite, 2020; Pintore et al, 2020)
For our experiment we choose three subsets of the data: i) 3D point cloud of an indoor area, which served as a base for 3D modeling of an input indoor model, ii) RGB images providing additional texture information for this area, with an association to the corresponding building spaces, iii) RGB images belonging to other buildings, used for transfer learning by Mask R-CNN
This paper presents an innovative approach to the automatic upgrade of existing indoor 3D models using up-to-date semantic information extracted from single RGB images with unknown camera poses
Summary
Building Information Modeling (BIM) is a diverse and multidisciplinary research subject with steadily increasing interest and demand (Czerniawski and Leite, 2020; Pintore et al, 2020). We could observe that the scope of BIMrelated topics is not anymore limited to purely professional usage More frequently it leaves space for various consumer products that rely on realistic 3D models created from spatial data. Detection of wall openings in indoor scenes may be a relatively complex task due to the existence of other objects, like pieces of furniture, that cause occlusions. The very complex and frequently changing environment of indoor spaces triggers the need for techniques that allow for an automatic update of existing 3D models, which increases their semantic coverage without the need for costly data capturing and repetitive reconstruction process
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
