On January 30, Herndon, VA-based BlackSky Technology Inc. announced that it won a contract from the Intelligence Advanced Research Projects Activity (IARPA) to move forward into the third and final phase of the multi-year Space-based Machine Automated Recognition Technique (SMART) program. Phase III emphasizes the transfer of SMART broad area search technologies into other U.S. government programs.
“BlackSky built a high-performance machine learning operations (MLOps) architecture for the SMART program that was originally designed to manage the analysis of one million square kilometers of land,” said Patrick O’Neil, BlackSky chief technology officer. “That foundational technology is now generating value elsewhere on other U.S. government search and discovery efforts, where we are analyzing more than 30 million square kilometers – more than 20 percent – of the Earth’s surface every month.”
Focused over large areas of the Earth’s surface, broad area search and discovery techniques help intelligence organizations get a wide-angle, baseline picture of aggregate change over time. For the SMART Program, all participants’ artificial intelligence (AI) change detection tools were deployed and evaluated for performance, cost and scalability within BlackSky’s MLOps architecture.
“BlackSky is using AI to reshape the broad area search mission with industry-leading speed and economics,” said O’Neil. “We have taken the lessons learned from SMART Phases I and II, folded them back into BlackSky Spectra, and quickly proven these investments can excel in an operational setting.”
With the core technology developed under the SMART program, BlackSky’s AI tools have detected and classified more than two million change observations with a high degree of accuracy over 120 thousand distinct sites after having analyzed five years of historic broad area satellite imagery from government and commercial sources.
Source: BlackSky
Start 2024 ahead of the competition with a paid subscription to IC News. You’ll get full access to our searchable archive of 10,000+ articles, plus new articles each weekday.
By 
