Geospatial ML with Amazon SageMaker (Preview)
Build, train, and deploy ML models faster using geospatial data
Up to 10 GB of free storage
for 30 days with the AWS Free Tier
Access readily available geospatial data sources, including satellite imagery, maps, and location data.
Efficiently process or enrich large-scale geospatial datasets with purpose-built operations such as resampling, mosaicking, and reverse geocoding.
Accelerate model building by using built-in, pretrained deep neural network models such as land cover segmentation and cloud masking.
Analyze geospatial data and explore model predictions on an interactive map using 3D accelerated graphics with built-in visualization tools.
How it works
The diagram has six sections. The first section has an image representing data and is titled, “Geospatial ML with Amazon SageMaker.” In this step, you can build, train, and deploy ML models using geospatial data.
The second section has four small images representing data sources and is titled, “Access data sources.” In this step, you can access geospatial data sources, from satellite imagery to mapping and location data, or upload your proprietary data.
The third section has an image of a globe with a small gear overlaying it and is titled, “Transform and enrich data.” In this step, you can use built-in operations such as reverse geocoding, mosaicking, and data labeling to prepare your data.
The fourth section includes an image representing models, with a small computer arrow clicking on them, and is titled, “Select or train a model.” In this step, you can select state-of-the-art, purpose-built models or bring and train your own model.
The fifth section has an image of a rocket flying out of a box and is titled “Deploy a model.” In this step, you can deploy and host your model to make predictions using real-world data.
The sixth and final section includes an image of a magnifying glass over a map and is titled, “Visualize predictions.” In this step, you can explore and analyze model predictions on a map from your Amazon SageMaker Studio notebook
Use cases
Assess risk and insurance claims
Measure risk, validate claims and prevent fraud, analyze damage impact from natural disasters on local economies, and track construction projects.
Inform trading strategies
Monitor financial assets globally, forecast market commodity prices, enhance your hedging or trading strategies, and mitigate the impact of price volatility.
Monitor climate change
Track deforestation and biodiversity, measure methane gas emissions, create climate resiliency plans, and improve power grid reliability.
Support sustainable urban development
Design more sustainable and livable urban environments, identify areas for land development, track traffic trends, or evaluate the feasibility of energy projects.
Maximize harvest yield and food security
View satellite images to diagnose plant health, insure and classify crops, predict harvest yield, forecast demand for agriculture produce, or detect farm boundaries.
Predict retail demand
Track high-growth city areas to improve sales or supply distribution channels, or combine location data with competitive intelligence to choose new store locations.
How to get started
AWS News Blog
Use Amazon SageMaker to build, train, and deploy ML models using geospatial data.
Developer Guide
Learn more about Amazon SageMaker geospatial capabilities in this step-by-step guide.
Example notebook
Learn how farmers can optimize crop production through advanced analytics and ML.