Program Summary
Adversaries often take photos and videos to claim responsibility for events or to illustrate capabilities. This media is sometimes confiscated by the DoD from a variety of devices, including laptops, cellphone cameras and memory cards. The volume of this visual media is quickly outpacing our ability to review, let alone analyze the contents of every image.
The question asked by warfighters and their analysts is how best can they turn adversary-captured ad hoc photos and videos into true “visual” intelligence. DARPA’s Visual Media Reasoning (VMR) program seeks to do just that by providing a software system that lets users ask queries of photo content, such as “What make and model of vehicle is that?” or “Is this person on our terrorist watch list?” or “Where is this building located?” If successful, VMR technology will serve as a force multiplier by extracting relevant information for human analysts and alerting them to scenes that warrant an analyst’s expert attention. The result of VMR may be an enhanced capability to generate intelligence required for successful counterinsurgency and counterterrorist operations.