The BIMiB project, funded by the German Federal Ministry of Education and Research under the “IngenieurNachwuchs – Kooperative Promotion” program (grant 13FH554IX6), ran from 1 May 2019 to 31 October 2023. It was coordinated by Prof. Dr. Dominic Becking of Hochschule Bielefeld and administered through the VDI Technology Centre in Düsseldorf. The consortium combined academic expertise with industrial partners, notably HOCHTIEF, ViCon GmbH and Pape Architekten AG, to develop automated methods for capturing and analysing structural systems in existing buildings. The overarching aim was to integrate the resulting data into standard BIM software, thereby facilitating adaptive reuse, reducing costs, and improving the economic viability and safety of building life‑cycle extensions.

Technically, the project was organised around four interrelated research strands. The first strand focused on machine‑learning‑based Scan2BIM. Researchers refined neural‑network architectures to process increasingly detailed and dense scan data, enabling the automated generation of BIM models that capture complex structural geometries with higher fidelity. The second strand integrated cognitive science insights into AI workflows, modelling human perception and logical reasoning to optimise the modelling pipeline and produce more realistic results. The third strand delivered a specialised processing pipeline that fuses 2‑D floor plans with 3‑D laser scans. By applying domain‑specific heuristics, the pipeline reduces errors and improves the quality of the merged data, providing a robust foundation for subsequent BIM construction. The fourth strand carried out evaluation trials on real building datasets and scans, confirming the practical applicability of the developed methods and demonstrating their potential to streamline BIM workflows in the construction industry.

A key scientific contribution was the creation of a metamodelling framework that incorporates building‑historical, structural and normative contexts. This framework allows machine‑readable knowledge representations to be embedded within the Scan2BIM process, enhancing the semantic richness of the resulting BIM models. Additionally, the project produced a catalogue of structural elements aligned with the IFC standard, detailing properties such as cross‑section, material and construction type, and distinguishing load‑bearing from non‑load‑bearing components. These artefacts support automated classification and validation of structural systems within BIM environments.

The evaluation phase involved rigorous testing against real‑world data. The trials showed that the integrated pipeline could reliably reconstruct structural elements from combined 2‑D and 3‑D sources, achieving a high degree of correspondence with manual measurements. While the report does not provide explicit numerical accuracy figures, it reports that the methods consistently outperformed baseline approaches in terms of completeness and precision, thereby validating the research hypotheses.

Collaboration was central to the project’s success. The academic partners at Hochschule Bielefeld and the VDI Technology Centre provided the research infrastructure and theoretical foundations, while the industry partners supplied real building data, practical constraints and validation scenarios. Regular workshops and joint development sessions ensured that the solutions remained aligned with industry needs and could be transferred to practice. The partnership model fostered an environment of open information exchange, mutual respect and continuous feedback, which the project team identified as a key factor for the positive outcomes and the potential for future joint ventures, including spin‑offs.

In summary, the BIMiB project achieved significant advances in automated structural system capture, data integration, and semantic modelling for existing buildings. By combining machine learning, cognitive insights, and a robust processing pipeline, it produced BIM models that are more accurate, contextually rich and ready for reuse. The close collaboration between academia and industry, supported by BMBF funding, enabled the translation of research findings into practical tools that can accelerate sustainable building life‑cycle extension and modernisation.