La NGA busca ayuda para entrenar a la IA para traducir imágenes con el fin de obtener información de inteligencia

La Agencia Nacional de Inteligencia Geoespacial (NGA) lanzó hoy un llamado a la industria —por un valor de hasta 708 millones de dólares durante un máximo de siete años— para obtener ayuda para entrenar sistemas de visión artificial impulsados por IA para, entre otras tareas, procesar imágenes satelitales e identificar objetivos de interés. NGA se hizo cargo de Maven del Departamento de Defensa en 2022. NGA recopila imágenes de satélites y aeronaves, las analiza y luego difunde los productos de inteligencia geoespacial (GEOINT) resultantes (como mapas 3D) a usuarios de todo el gobierno de EE. UU., incluidos líderes del Departamento de Defensa y comandantes militares.


WASHINGTON — The National Geospatial Intelligence Agency (NGA) today issued a call to industry — worth up to $708 million over a maximum of seven years — for help training AI-driven computer vision systems to, among other tasks, process satellite imagery and identify targets of interest.

Under the Sequoia program indefinite delivery/indefinite quantity (ID/IQ) contract, chosen vendors will provide data labeling, which allows artificial intelligence and machine learning systems to discriminate among objects. It is a foundational capability in particular for NGA’s sprawling Maven program, according to NGA officials.

NGA took over Maven from the Defense Department in 2022. NGA gathers imagery from satellites and aircraft, analyzes it, and then disseminates the resultant geospatial intelligence (GEOINT) products (such as 3D maps) to users across the US government, including to DoD leaders and military commanders.

“For example, the ID/IQ contract will support the NGA Maven Program, which applies AI capabilities to intelligence, surveillance and reconnaissance sensors and platforms, primarily through computer vision,” the NGA announcement said. “The NGA Maven CV [computer vision] algorithms, which rely heavily on robust data labeling, can perform a number of automated tasks, including but not limited to object detection, object tracking, object classification and pattern detection.”

The Sequoia contract also will support “natural language processing, analytic models and AI/ML models that support business process automation for the GEOINT mission,” the announcement added.

Vice Adm. Frank Whitworth, NGA director, told reporters on Aug. 30 that the call is “the largest data labeling request for proposal in the US government,” and “represents a significant investment in computer vision, machine learning and AI.”

The increasing amount of GEOINT data pouring in from myriad new satellites and other sensor platforms is posing a challenge for the Intelligence Community. While AI systems will help speed analysis, AI visual models first must be taught to accurately recognize military targets and ferret out abnormal activities.

Fuente: https://breakingdefense.com