Buildings footprint segmentation
Object Segmentation | Satelite Imagery | Sentinal Dataset | Computer Vision | Deep Learning
- Developed a UNet-based model to segment building structures from
multi-spectral Sentinel imagery. - Led the model training by successfully applying AI to geospatial data, achieving a
precision of 92%. - Conducted performance evaluation and validation, achieving robust results and demonstrating scalability.
Semantic segmentation of buildings footprint. We leveraged a datasets of Massachusetts region.