Towards fully automatic Scan-to-BIM: A prototype method integrating deep neural networks and architectonic grammar

Abstract: Building Information Modeling (BIM) has presented great potential in the construction industry. Scan-to-BIM is highly demanded to boost automation for creating as-built BIMs. This paper focuses on an extreme case — fully automatic Scan-to-BIM, on which recent advances in deep neural networks (DNNs) shed light. We present a prototype FLKPP that integrates DNNs with architectonic grammar. FLKPP won 2nd place in 3D reconstruction and 3rd place in 2D reconstruction in the 2nd International Scan-to-BIM Challenge. Nevertheless, the results of all methods in the Challenge were still limited, indicating a long way leading to the complete automation of Scan-to-BIM.