Project Maven: computer vision joins intelligence work
On 26 April 2017 Deputy Secretary of Defense Robert O. Work signed a memo establishing the Algorithmic Warfare Cross-Functional Team "to accelerate DoD's integration of big data and machine learning". Its first task was to automate the processing of full-motion video from drones for the campaign against ISIS.
Why it matters
The moment object recognition in video stopped being a task on a dataset and became a standing link in the intelligence cycle. Within a year this same programme would cause the first serious rupture between a laboratory and its own engineers.
The memo sets the team's objective as turning the enormous volume of data available to the department "into actionable intelligence and insights at speed". It was to begin with computer vision algorithms for object detection, classification and alerts in drone video, work in 90-day sprints and then move into other areas of defence intelligence. The Under Secretary of Defense for Intelligence was to oversee it and his Director for Defense Intelligence (Warfighter Support) to serve as its director, reporting directly to the Deputy Secretary. The memo names no sum; per the secondary source, initial funding was 70 million dollars, and Jack Shanahan was director from April 2017 to December 2018. Per the same source, Maven's first real use came in 2017 in East Africa, where special forces were operating against al-Shabaab, and in early December 2017 the system was fielded to defence intelligence analysts to support drone missions. The National Geospatial-Intelligence Agency says the GEOINT aspects of Maven were entrusted to it in January 2023; the programme became a Program of Record on 7 November 2023, according to the press.