Your YouTube videos could soon be scanned and evaluated for terror threats, thanks to a new project funded by the US intelligence community that'll create a searchable warehouse of open-source clips.
Iarpa, the spy agency's out-there research arm, is behind the program, called Automated Low-Level Analysis and Description of Diverse Intelligence Video (ALADDIN).
The advent of mobile phone cams and online video hubs means thousands of clips are uploaded every day. And while the quality of UAV or spy-cam feeds is relatively consistent, and the videos usually contain similar imagery, "the unconstrained video clips produced by anyone who has a digital camera present a significant challenge," Iarpa notes.
Uploaded videos contain such diverse scenes and situations, not to mention grainy images and sound, that it's much harder to prep algorithms for automated evaluation. And human analysts only have so much time for the "eyes-on-video/ears-on-audio" routine.
In ALADDIN, Iarpa wants a boundless database of open source video clips, and the ability to search "for the occurrence of specific events of interest". Once the system finds the relevant videos, it should be able to "rapidly and automatically produce a textual English-language recounting…describing the particular scene, actors, objects and activities involved."
Of course, despite the challenges of analysing uploaded videos, spy agencies are probably already doing it. In 2008, the director of the Office of the Director of National Intelligence's Open Source centre noted that "YouTube…carries some unique and honest-to-goodness intelligence."
Wired.com has been expanding the hive mind with technology, science and geek culture news since 1995.