Amazon SageMaker FAQs for Geospatial ML
Q: What is geospatial data?
Geospatial data represents features or objects on the earth’s surface. The first type of geospatial data is vector data, which uses two-dimensional geometry such as points, lines, or polygons to represent objects such as roads and land boundaries. Geo-tagged location data is also considered to be vector data. It includes points of interest—for example, the Eiffel Tower—location-tagged social media posts, latitude and longitude coordinates, or different styles and formats of street addresses. The second type of geospatial data is raster data, such as imagery collected by satellites, aerial platforms, or remote sensing platforms. This data type uses a matrix of pixels to define where features are located. You can use raster formats for storing data that varies.
Q: How do I get geospatial data?
Amazon SageMaker geospatial capabilities allow you to use geospatial data, such as Landsat 8 and Sentinel-2. You can also import your own data—including location data generated from GPS devices, connected vehicles or Internet of Things (IoT) sensors, retail store foot traffic, geomarketing and census data, or data acquired from third-party vendors. SageMaker geospatial capabilities also help you enrich this data using purpose-built functions from Amazon Location Service, such as converting latitude and longitude locations to street addresses.
Q: What are SageMaker geospatial capabilities?
SageMaker geospatial capabilities make it easier for data scientists and machine learning (ML) engineers to build, train, and deploy ML models for making predictions using geospatial data. You can bring your own data, such as Planet Labs satellite data from Amazon Simple Storage Service (Amazon S3), or acquire data from Open Data on AWS, Amazon Location Service, and other SageMaker geospatial data sources.
Q: How can I enhance efficiency with SageMaker geospatial capabilities?
SageMaker geospatial capabilities provide users with instance types and notebooks optimized for geospatial ML. These notebooks have embedded visualization tools and commonly used open-source geospatial libraries, as well as purpose-built models, algorithms, and functions. You can simplify your data preprocessing with built-in geospatial operations such as map matching. Speed up geospatial ML model development and lower total cost of ownership by using one of the prebuilt models, or develop your own geospatial ML models. You can visualize predictions layered on a map with the built-in visualization tools, which allow faster collaboration.
Q: Why should I use SageMaker geospatial ML capabilities?
You can use SageMaker geospatial ML capabilities to make predictions on geospatial data faster than do-it-yourself solutions. SageMaker geospatial capabilities make it easier to access geospatial data from your existing data lakes, open-source datasets, and other SageMaker geospatial data sources. SageMaker geospatial capabilities minimize the need for building custom infrastructure and data preprocessing functions by offering purpose-built algorithms for efficient data preparation, model training, and inference. You can also create and share custom visualizations and data with your organization from Amazon SageMaker Studio. SageMaker geospatial capabilities include pretrained models for common uses in agriculture, real estate, insurance, and financial services.