📄️ Basic Semantic Segmentation
A complete example showing how to train a U-Net model for binary segmentation using a custom dataset.
📄️ Multispectral Satellite Imagery Segmentation
A complete example showing how to train a U-Net model for building and vegetation segmentation on 4-band RGBI (Red, Green, Blue, Near-Infrared) satellite imagery stored as GeoTIFF files.
📄️ Multi-Class Semantic Segmentation
A complete example showing how to train a model for land cover classification with multiple mutually-exclusive classes such as buildings, vegetation, roads, and water.
📄️ Building Extraction with Frame Field Models
A complete example showing how to train a Frame Field segmentation model for precise building footprint extraction from aerial imagery, with post-processing to produce clean polygon vectors.
📄️ Object Detection with Faster R-CNN
A complete example showing how to train a Faster R-CNN model to detect buildings or vehicles in aerial imagery, using bounding box annotations stored in JSON files.