Lockheed Martin’s new Global Automated Target Recognition tool showed high accuracy rate during tests ahead of its launch at the GEOINT 2019 symposium in San Antonio.
The company said GATR runs in the cloud and collects satellite imagery from open-source deep learning libraries, which enables the tool to speed up process to identify objects in large areas.
“There’s more commercial satellite data than ever available today, and up until now, identifying objects has been a largely manual process,” said Maria Demaree, vice president and general manager of Space Mission Solutions at Lockheed. “Artificial Intelligence models like GATR keep analysts in control while letting them focus on higher-level tasks.”
During tests, GATR took two hours to find fracking sites across the entire state of Pennsylvania and Lockheed achieved an accuracy rate of over 90 percent on test models.
The company said it utilized Maxar Technologies’s Geospatial Big Data platform to speed up training of algorithms to recognize targets by feeding GATR with 100 petabyte satellite imagery and millions of data labels.