National Geospatial-Intelligence Agency (NGA) Director of Analysis Sue Kalweit called on audience members at the GEOINT Symposium Government Pavilion Tuesday to help the agency move into an era of analytic modernization, specifically in automation.
“We need you as an industry partner to help us understand how to take advantage of using all the data,” Kalweit said alongside a panel of NGA analysts. “The current and future state of technology puts the power of integrating imagery intelligence and geospatial information on steroids.”
GEOINT is more than imagery, Kalweit said. It’s “all spatially and temporally referenced data, from any and every source.” She said with exponentially mounting data, NGA needs help from industry to automate analysis. Every week, 10 percent of an analyst’s time is spent just collecting data.
“Most of our SOM (Structured Observation Management) collection right now is a manual process,” she said. “You’ve got to give my analysts their time back every week so they can do the higher-level analytics.”
Senior GEOINT Analyst Bryan Weaver said SOM, which brings GIS practices to the imagery analysis tradecraft, needs a data model that’s responsive to quickly changing customer needs. The classic example is, “Where are the tanks?” But the question, he said, may become, “What is the capability of the adversary, and are they about to invade?”
Kalweit explained what NGA is looking for in terms of computer vision. Change detection refers to shortening the time window on which analysts need to focus. Object identification means saying with confidence that an object is, for example, a tank or an aircraft. The next level is object characterization, such as identifying a vehicle specifically as a Ford F-150 with an extended cab. The ultimate goal—object contextualization—means you know the Ford F-150 with the extended cab is being used for pleasure, because the person behind the wheel never drives it to work. Achieving object contextualization, she said, is the “holy grail.”