Home News DigitalGlobe releases commercial version of GBDX Notebooks, accelerating machine learning innovation

DigitalGlobe releases commercial version of GBDX Notebooks, accelerating machine learning innovation

Monaco satellite imagery

US: DigitalGlobe, a Maxar Technologies company, has announced today the commercial release of GBDX( Geospatial Big Data) Notebooks, allowing customers to easily extract valuable insight they want from DigitalGlobe’s Geospatial Big Data platform, (GBDX), eliminating the heavy lifting that accompanies managing satellite imagery at scale.

GBDX Notebooks leverages the familiar open source Jupyter Notebooks environment to provide access to DigitalGlobe’s 18-year, 100 PB library of high-resolution, multispectral imagery, as well as pre-built machine learning and remote sensing algorithms. In addition to three e-commerce subscription levels, GBDX Notebooks offers a community tier, furnishing customers with ready access to the latest imagery from NASA’s Landsat satellite, ESA’s Sentinel satellite, DigitalGlobe’s Open Data Program, and DigitalGlobe’s complete IKONOS archive. DigitalGlobe’s IKONOS satellite, the world’s first commercial satellite launched in 1999, has a 15-year imagery library and it’s the only sub-meter satellite imagery accessible for the creation of open source algorithms. All algorithms developed within the community tier are available as open source.

Using one of the three available e-commerce tiers, customers can develop and run at scale proprietary algorithms, Maxar-developed algorithms, or open source algorithms, including:

  1. Mapping Navigable Airspace for Drones: By combining DigitalGlobe’s imagery, digital surface models, a tree detection model, and Ecopia Building Footprints powered by DigitalGlobe, this GBDX Notebook visualizes obstructions that would impact the flight paths of drones and UAVs, solving a key problem for the emerging drone and UAV navigation sectors.
  2. Assessing Forest Fire Risk to Buildings: Calculating by hand the fire risk for every building in a city would be a time-consuming task. This GBDX Notebook combines high-resolution satellite imagery, building footprints, Normalized Difference Vegetation Index (NDVI) to determine plant health, a tree detection model, and defensible space standards to quickly identify which buildings would have a high risk of burning in the event of a wildfire, helping a city determine where to focus its fire mitigation efforts.
  3. Change Detection for Boat Traffic: This GBDX Notebook combines feature extraction, change detection over time, and open source weather data to find business insights from imagery, specifically helping businesses identify when to expect more customers.

“DigitalGlobe is committed to helping customers unlock valuable insights from our powerful data by applying machine learning to solve complex problems facing businesses, including the insurance, automotive, energy, and technology industries,” said Amy Minnick, DigitalGlobe Senior Vice President and General Manager of Commercial.

“DigitalGlobe developed GBDX Notebooks to encourage more people, like developers, data scientists, imagery scientists, and machine learning and deep learning researchers, to extract location intelligence from satellite imagery, enabling commercial customers to make business decisions with confidence, faster, and at scale,” said Dr. Shay Har-Noy, DigitalGlobe Vice President of Unified Platform. “By gifting the IKONOS archive to science, we anticipate wider adoption of satellite imagery as a foundational data layer within the machine learning community and with customers.”

GBDX Notebooks is a new and powerful component of a comprehensive machine learning platform for satellite imagery providing training data, an imagery chipping service, labeling tools, and high-speed imagery inference. Maxar’s DigitalGlobe has released a series of GBDX Notebooks tutorials. Read the first tutorial here, and for opportunities to participate in early access to future machine learning capabilities, sign up here.

GBDX Notebooks will also be available as a demo in DigitalGlobe’s booth at the Locate Conference in San Francisco on May 30–31, 2018.