A Tutorial on GeoAI: Designing Footprint Extraction from NAIP Imagery Using U-Net, Grounding DINO, SAM, and Mask R-CNN

In this tutorial, we design a complete GeoAI workflow for extracting building footprints from high-resolution NAIP aerial imagery.

We begin by configuring the geospatial deep learning environment, downloading raster imagery and vector labels, and inspecting their spatial properties before generating georeferenced image chips and segmentation masks.