Deep Residual U-Net for Building Footprint Extraction
A reproducible PyTorch pipeline trained on the SpaceNet Rio de Janeiro dataset for extracting building footprints from high-resolution satellite imagery.
Projects
DeepSlide: Landslide Detection & Segmentation
An application for automated landslide detection from multi-source satellite imagery using 13+ deep learning models.
Tree / Orchard Detection & Counting from High-Resolution Satellite Images
An end-to-end density-map regression U-Net pipeline to detect and count individual trees and orchards from high-resolution satellite imagery.