LoD2 Building Reconstruction with Geospatial AI

Mar 4, 2026·
Dr. Shabnam Jabari, Associate Professor, P.Eng
Dr. Shabnam Jabari, Associate Professor, P.Eng
Faezeh Soleimani Vostikolaei
Faezeh Soleimani Vostikolaei
· 1 min read

ASIL develops AI-assisted methods for LoD2 building reconstruction using orthophotos, point clouds, and mobile mapping imagery.

Focus

  • Extract roof components using graph-based and deep learning approaches
  • Detect facade elements under occlusion, complex viewpoints, and varying image conditions
  • Generate robust building models for large urban areas

Research Outputs

Recent work includes journal and conference publications on multimodal segmentation, feature fusion, and reconstruction quality assessment.

LoD2 Deep Learning Building Modeling
Dr. Shabnam Jabari, Associate Professor, P.Eng
Authors
Dr. Shabnam Jabari, Associate Professor, P.Eng
Principal Investigator and Lab Director
Associate Professor at UNB working on geospatial AI, remote sensing, and urban digital twins for climate resilience and disaster management.
Faezeh Soleimani Vostikolaei
Authors
Faezeh Soleimani Vostikolaei
Postdoctoral Fellow • Project Technical Leader
Postdoctoral researcher leading ASIL’s LoD2 building modeling pipeline, with focus on bimodal segmentation and graph neural networks for roof component extraction.