# torchgeo/torchgeo

TorchGeo: datasets, samplers, transforms, and pre-trained models for geospatial data

Repository: https://github.com/torchgeo/torchgeo
Canonical: https://ross.abutalabs.com/products/torchgeo
Homepage: https://torchgeo.org/
Language: Python
License: MIT
License Family: permissive
Topics: pytorch, torchvision, datasets, models, transforms, remote-sensing, deep-learning, earth-observation, computer-vision, geospatial, satellite-imagery
Last push: 2026-08-26T17:14:13+00:00

## Health v2 (maintenance only)
Score: 95/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 86, longevity 100
- inputs: {"age_days": 1930, "days_push": 7, "days_rel": 19, "gap_med": 72.5, "n_releases_24m": 9}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4159, forks 580 (observed 2026-08-28T04:08:37.434959+00:00)

## What it is
TorchGeo is a PyTorch domain library, similar to torchvision, providing datasets, samplers, transforms, and pre-trained models specific to geospatial data. It makes it simple for machine learning experts to work with geospatial data and for remote sensing experts to explore machine learning solutions.

## Use cases
- train deep learning models on satellite imagery
- load geospatial datasets in PyTorch
- fine-tune pretrained models on Sentinel-2 multispectral data
- do land cover classification with remote sensing data
- semantic segmentation of aerial imagery
- handle CRS-aligned sampling of geospatial rasters
- work with multispectral bands without RGB workarounds
- benchmark models on earth observation data

## When to choose
- you want torchvision-style workflows for satellite or aerial imagery
- you need CRS-aware datasets, samplers, and multispectral transforms out of the box
- you want pretrained remote sensing weights like SeCo, MoCo, MAE, or DOFA
- you use PyTorch or Lightning and need geospatial data pipelines

## When to avoid
- your project is not Python/PyTorch based
- you only need GIS processing without machine learning
- you need a general-purpose computer vision library without geospatial metadata support

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, image-processing, data-science
- domain: machine-learning, deep-learning, computer-vision
- platform: python, cross-platform
- tags: pytorch, torchvision, satellite-imagery, remote-sensing, earth-observation, multispectral, pretrained-models, lightning, crs-aware, datasets, samplers, transforms, geospatial, gpu

## Member repositories
- torchgeo/torchgeo (main) score 95

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:37.434959+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:22:50.160823+00:00, confidence not recorded.
  - readme: https://github.com/torchgeo/torchgeo (fetched 2026-08-28T04:08:37.434959+00:00, sha e8bea9316613)
  - homepage: https://torchgeo.org/ (fetched 2026-08-29T09:13:39.742955+00:00, sha 1e68e24088f9)
  - site_page: https://docs.torchgeo.org/en/stable/tutorials/getting_started.html (fetched 2026-08-29T09:13:39.752087+00:00, sha d57a8b2ad72a)
  - site_page: https://docs.torchgeo.org/en/stable (fetched 2026-08-29T09:13:39.753958+00:00, sha 52bb18aa789b)
  - registry_pypi: https://pypi.org/pypi/torchgeo/json (fetched 2026-08-29T09:13:39.755731+00:00, sha 02147c712875)
- Data as of 2026-08-30T08:39:29.467469+00:00.