Abstract
In circular construction, sourcing reclaimed materials for design projects is often hindered by inconsistent data formats and lack of 3D data integra tion. This study addresses these barriers by proposing a digital workflow to stream line information flow between 3D architectural models and circular materials. Our approach includes three key elements: a PostgreSQL database integrated into the CAD environment for easy querying of reclaimed materials, an automated 3D model generation system using image-segmentation and depth-estimation models with OpenCV andMETA Segment AnythingModel, and an intuitive user-interface within 3D CAD software. The workflow detects and segments elements like windows and doors from image data, performs perspective transformations, and converts the data into JSON format for 3D model generation in CAD software to reduce manual modeling effort. It also includes a database for standardized material data and an intuitive user interface integrated into 3D modeling software to provide architects real-time access to circular materials. The interface features cost calculations, greenhouse gas emissions visualizations, and summary tools, ensuring digital models correlate with physical materials. This user-friendly approach simplifies incorporating reclaimed materials into designs, paving the way for future investigations into more complex shapes and materials. This integration significantly enhances efficiency for designers and architects during the design stage. Overall, in this research, we aim to showcase a more user-friendly approach to simplify the use of reclaimed materials. The proposed workflow enables architects and designers to efficiently and effectively incorporate reclaimed materials with detailed characteristics into their designs. Future work will expand the database to include diverse material types, such as structural components, and address the challenges of photo-based 3D model generation for more complex geometric shapes. Show more
Publication status
publishedExternal links
Book title
Scalable Disruptors: Design Modelling Symposium Kassel 2024Pages / Article No.
Publisher
SpringerEvent
Subject
Computer Vision; Machine Learning; DatabaseOrganisational unit
09750 - De Wolf, Catherine / De Wolf, Catherine
09566 - Dillenburger, Benjamin / Dillenburger, Benjamin
09724 - Langenberg, Silke / Langenberg, Silke
02284 - NFS Digitale Fabrikation / NCCR Digital Fabrication
Funding
-- - NCCR Digital Fabrication (SNF)
More
Show all metadata
ETH Bibliography
yes
Altmetrics