DARPA develops new tools to help process video data
Effort will help analysts detect activities and threats more quickly and easily
The constellation of ground-based and airborne sensors deployed by U.S. forces in Southwest Asia has been both a blessing and a curse. Although it allows warfighters to track and identify activities such as the planting of improvised explosive devices, the sheer volume of incoming data often works against them.
To control the information fire hose that wastes both manpower time and resources, the Defense Advanced Research Projects Agency has launched two related programs to provide warfighters with automated tools to quickly sift through large volumes of data. The Image Retrieval and Analysis Tool (VIRAT) and the Persistent Stare Exploitation and Analysis System (PerSEAS) are designed to identify and tag certain types of actions in raw video streams to bring them to a human operator’s attention.
“Bad guys do bad things, such as all the actions involved in burying an IED – so it is activity that matters," VIRAT and PerSEAS program manager Mita Desai said in a DARPA statement. "This is especially true when the bad guys look, dress and drive vehicles like those around them.”
VIRAT is focused on full-motion video from platforms such as unmanned aerial systems and aerostats. It works in both forensic mode to identify activities in thousands of hours of archived, untagged data and in streaming mode. VIRAT focuses on identifying activities that are short in duration and take place in a small area. Algorithms from VIRAT support some of the underlying functions of PerSEAS.
PerSEAS is designed to observe multiple actions over a long period of time across a large geographic area to identify dangerous combinations and patterns of actions. This system will analyze data from platforms such as Constant Hawk and Gorgon Stare. According to DARPA, PerSEAS is undergoing additional research and development work before it is deployed for field testing.
“The objectives of VIRAT and PerSEAS are not to replace human analysts, but to make them more effective and efficient by reducing their cognitive load and enabling them to search for activities and threats quickly and easily,” Desai said.
No comments:
Post a Comment