H-BIM models generation with artificial intelligence

The data of our cultural heritage obtained through three-dimensional surveys represent complex unstructured geometric structures. The introduction of automatic methods of semantic segmentation and hierarchy between the parts would allow the understanding and use of this data in asset protection operations. In this context, the contribution presents a semi-automatic workflow for the three-dimensional reconstruction of H-BIM models, exploiting a hierarchical classification approach (MLMR) for the optimization of the results.