Advancing Earth Observation Through Machine Learning: A TorchGeo Tutorial

TL;DR

This paper introduces TorchGeo, a PyTorch library tailored for geospatial data, and demonstrates its application through a tutorial on multispectral water segmentation from Sentinel-2 imagery, highlighting its utility in Earth observation tasks.

Contribution

The paper presents a comprehensive tutorial showcasing TorchGeo's core abstractions and an end-to-end case study for multispectral water segmentation, facilitating geospatial machine learning workflows.

Findings

Effective segmentation of water bodies using Sentinel-2 data

Demonstration of training and applying models with TorchGeo

Generation of GeoTIFF outputs for geospatial analysis

Abstract

Earth observation machine learning pipelines differ fundamentally from standard computer vision workflows. Imagery is typically delivered as large, georeferenced scenes, labels may be raster masks or vector geometries in distinct coordinate reference systems, and both training and evaluation often require spatially aware sampling and splitting strategies. TorchGeo is a PyTorch-based domain library that provides datasets, samplers, transforms and pre-trained models with the goal of making it easy to use geospatial data in machine learning pipelines. In this paper, we introduce a tutorial that demonstrates 1.) the core TorchGeo abstractions through code examples, and 2.) an end-to-end case study on multispectral water segmentation from Sentinel-2 imagery using the Earth Surface Water dataset. This demonstrates how to train a semantic segmentation model using TorchGeo datasets, apply the model to a Sentinel-2 scene over Rio de Janeiro, Brazil, and save the resulting predictions as a GeoTIFF for further geospatial analysis. The tutorial code itself is distributed as two Python notebooks: https://torchgeo.readthedocs.io/en/stable/tutorials/torchgeo.html and https://torchgeo.readthedocs.io/en/stable/tutorials/earth_surface_water.html.