If a picture is worth a thousand words, the scientists at the Defense Advanced Research Projects Agency (DARPA) would like to make that about a billion with a new software intelligent program.
DARPA this month said it will detail a new system it would like to see built known as the Visual Media Reasoning (VMR) program. The main idea is to develop an advanced software program that can "turn 'dumb' unstructured, ad hoc photos and video into true visual intelligence."
DARPA says insurgents and terrorist groups frequently use video, still and cell phone cameras to document their training and operations and occasionally post this content to widely available websites.
"The volume of this visual media is growing rapidly and is quickly outpacing our ability to review, let alone analyze, the contents of every image. The VMR technology will extract tactically relevant information for the human analyst and alerting the analyst to scenes that warrant the analyst's expert attention. VMR will be an enhanced capability to generate the intelligence required for successful counter-insurgency and counter-terrorism operations" the agency said.
DARPA said it expects one of the primary technical achievements of the VMR system will be the creation of an intelligent meta-algorithm- a system controller that interprets the user's submitted image, pre-processes it as necessary (looks for meta-data, recognizes text, sharpens, etc.), decides which computer vision algorithms to apply for which sub-tasks and in what sequence, matches the appropriate algorithms and their results to available datasets, and acquires user input to aid the process as needed.
"While the VMR program will focus primarily on creating a software system that leverages the existing and rapidly growing corpus of computer vision science, there will be support for a small amount of fundamental research into technologies that enable the VMR concept such as new approaches to image processing, pattern matching, and database indexing with respect to the image attributes of persons, objects, places, and time," DARPA stated. "The computer vision/image understanding field has matured greatly over the last 40 years and is currently seeing rapid new advances driven by private sector trends such as augmented reality apps on smart phones and the integration of powerful image search capabilities into the popular internet search engines. Rather than recreate the wheel, VMR will exploit these current and emerging advances by employing ensembles of best-of-breed image recognition techniques, from 3D object extraction to color signatures to auto-indexing of image content and more."
DARPA will hold an Industry Day briefing on VMR August 31 in Arlington, VA. A Broad Agency Announcement solicitation should follow.
Follow Michael Cooney on Twitter: nwwlayer8
Layer 8 Extra
Check out these other hot stories: