Semantic Enrichment of a BIM Model Using Revit: Automatic Annotation of Doors in High-Rise Residential Building Models Using Machine Learning

This study explores the potential of automated fire safety conformity checks using a BIM tool. The focus is on travel distance regulations, one of the major concerns in building design. Checking travel distances requires information about the location of exits. Preferably, the Building Information Model (BIM) of the building should contain such information, and if not, user input can be requested. However, a faster, yet still reliable and accurate, methodology is strived for. Therefore, this study presents an automated solution that uses machine learning to add the required semantics to the building model. Three algorithms (Bagged KNN, SVM, and XGBoost) are evaluated at a low Level of Detail (LOD) BIM models. With precision, recall, and F1 scores ranging from 0.87 to 0.90, the model exhibits reliable performance in the classification of doors. In a validation process with two separate sample buildings, the models accurately identified all ’Exits’ in the first building with 94 samples, with only 5 to 6 minor misclassifications. In the second building, all models- with the exception of the SVM - correctly classified every door. Despite their theoretical promise, oversampling techniques do not enhance the results, indicating their inherent limitations.