| id |
caadria2025_487 |
| authors |
Klimenka, Mikita, Stoddart, James, Villaggi, Lorenzo, Zhao, Dale, Rampini, Arianna, Gaier, Adam, Locke, John and Benjamin, David |
| year |
2025 |
| title |
Seeing Inside Buildings: Leveraging generative AI and multimodal data to automate building material audits |
| source |
Dagmar Reinhardt, Christiane M. Herr, Anastasia Globa, Jielin Chen, Taro ?Narahara, Nicolas Rogeau (eds.), ARCHITECTURAL INFORMATICS - Proceedings of the 30th CAADRIA Conference, Tokyo, 22-29 March 2025, Volume 3, pp. 539–548 |
| summary |
Buildings account for almost 40% of global carbon emissions, and one of the most immediate and scalable strategies to reduce these emissions is to reuse existing buildings rather than construct new ones as our needs for the built environment change and grow. However, building reuse and material upcycling is often hindered by the complexity of building material audits, which require expensive site visits. We present a Generative AI approach to predict the structural and material make-up of existing buildings from multimodal geospatial, technical, and cadastre data. Leveraging a dataset of 100 buildings across the United States with corresponding building 3D scans, geolocation, and construction data, we demonstrate the capability of a stable diffusion model to reliably predict structural diagrams for subsequent estimation of material contents. Additionally, we highlight the effectiveness of combining additional diverse data inputs – from cadastre reports to photographs and satellite views – to accurately predict interior structure. Our approach aims to allow building owners to bypass the need for complex inspections and expensive expert analysis and lay the foundation for ubiquitous and affordable building material audits. This process also offers designers actionable data about material reuse to streamline and accelerate circularity for existing building design. |
| keywords |
Material Reuse, Building Retrofits, Generative AI, Multimodal Data, Urban Data |
| series |
CAADRIA |
| full text |
 file.pdf (1,002,800 bytes) |
| last changed |
2025/03/21 12:09 |
|
