# azavea/raster-vision

An open source library and framework for deep learning on satellite and aerial imagery.

Repository: https://github.com/azavea/raster-vision
Canonical: https://ross.abutalabs.com/products/raster-vision
Homepage: https://docs.rastervision.io
Language: Python
License: Apache-2.0
License Family: permissive
Topics: deep-learning, computer-vision, remote-sensing, geospatial, object-detection, semantic-segmentation, classification, machine-learning, pytorch
Last push: 2026-06-04T14:59:54+00:00

## Health v2 (maintenance only)
Score: 61/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 85, release rhythm 8, longevity 100
- inputs: {"age_days": 3499, "days_push": 90, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2240, forks 397 (observed 2026-08-28T04:06:29.698531+00:00)

## What it is
Raster Vision is an open source Python library and low-code framework for building computer vision models on satellite, aerial, and other large georeferenced imagery sets. It provides built-in support for chip classification, object detection, and semantic segmentation with PyTorch backends, and orchestrates the full ML pipeline from training-data analysis through model bundling for deployment.

## Use cases
- train object detection models on satellite imagery
- run semantic segmentation on aerial or drone images
- classify land cover from remote sensing data
- build repeatable deep learning pipelines for large imagery sets
- process georeferenced raster data for machine learning
- scale geospatial ML experiments on AWS Batch or SageMaker
- make predictions on oblique drone imagery

## When to choose
- You need to train or apply CV models on georeferenced satellite, aerial, or drone imagery
- You want a low-code way to configure end-to-end pipelines: chipping, training, prediction, evaluation, and model bundling
- You need outputs written back in geo-referenced formats
- You want to run experiments repeatably in the cloud via AWS Batch or SageMaker
- You prefer a pip-installable library with utilities for reading geo-referenced data on top of PyTorch

## When to avoid
- Your task is general-purpose computer vision on ordinary photos with no geo-referencing
- You need CV tasks beyond classification, object detection, or semantic segmentation, such as instance segmentation or generative models
- You need a GUI-based annotation or labeling tool rather than a training pipeline
- You work primarily in a non-Python or non-PyTorch ecosystem
- You need real-time streaming inference rather than batch-oriented workflows

## Facets
- artifact type: framework
- maturity: active
- function: deep-learning, machine-learning, computer-vision, image-processing
- domain: deep-learning, computer-vision, machine-learning, data-science
- platform: python, cloud
- tags: geospatial, remote-sensing, satellite-imagery, aerial-imagery, pytorch, object-detection, semantic-segmentation, image-classification, chip-classification, raster-data, aws-batch, aws-sagemaker, gis, drone-imagery, low-code, docker, gpu

## Member repositories
- azavea/raster-vision (main) score 61

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:29.698531+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 2026-08-30T02:44:07.326861+00:00, confidence not recorded.
  - readme: https://github.com/azavea/raster-vision (fetched 2026-08-28T04:06:29.698531+00:00, sha d78484a5a731)
- Data as of 2026-08-30T08:39:29.467469+00:00.