BIMStruct3D: A Fully Automated Hybrid Learning Scan-to-BIM Pipeline with Integrated Topology Refinement

[Abstract] Automatic generation of Building Information Models (BIM) from building scans is a key challenge in architecture and construction. We present a modular pipeline for generating IFC-compliant BIM from 3D point clouds. The hybrid approach combines learning-based semantic segmentation with topology-aware geometric reconstruction to model structural elements accurately. We propose vIoU, adapting voxel-based overlap evaluation to Scan-to-BIM by enabling holistic, instance-matching-free comparison of reconstructed and ground-truth models. We release the German Hospital dataset (DeKH), including high-resolution point clouds, ground truth BIMs, and semantic annotations. Experiments on DeKH and CV4AEC datasets show significant improvements over a RANSAC-based baseline, demonstrating robustness and scalability.

The paper was presented at the 2026 European Conference on Computing in Construction (2026 EC³). The final paper and video presentation are available here. DOI: 10.35490/EC3.2026.