August 21, 2025Remote SensingOpen Access

Single Image to Semantic BIM: Domain-Adapted 3D Reconstruction and Annotations via Multi-Task Deep Learning

View Full Paper

Ask AI

Bookmark

Authors

Discussion

This research demonstrates automated 3D reconstruction and semantic labeling in built environments, suggesting enhancements in BIM applications.

Achieving high accuracy, the method enables accurate 3D mesh reconstruction from single RGB images.
Experimental evaluations show a reconstruction precision with RMSE as low as 0.02 and a semantic accuracy of 81.89%.
Employing a multi-task learning approach, the pipeline integrates depth estimation and semantic segmentation effectively.
The study addresses limitations in traditional scan-to-BIM workflows, supporting the rapid digitization of complex built environments.

Erişen et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

This research demonstrates automated 3D reconstruction and semantic labeling in built environments, suggesting enhancements in BIM applications.