The Intelligence Advanced Research Projects Activity has launched the Video Linking and Intelligence from Non-Collaborative Sensors program, which seeks to develop re-identification algorithms that would make it possible to autonomously associate content found across footage recorded by diverse video sensors and geo-locate that re-identified content.
According to Video LINCS Program Manager Reuven Meth, the resulting capabilities would provide a new tool that intelligence analysts can use to process and identify threats that might be recorded in large volumes of disparate video imagery, IARPA said Tuesday.
The Video LINCS program will unfold in three phases across 48 months. The first phase involves demonstrating the feasibility of the re-identification and geolocation of persons in video compilations. The second phase would cover vehicles. The third phase would cover generic objects without prior knowledge as to what kind of objects they would be.
Although the Video LINCS program is in pre-solicitation status, IARPA has already selected MITRE, MIT Lincoln Laboratory and the National Institute of Standards and Technology as testing and evaluation partners.
