Needed at IARPA: Better tools for drone surveillance

The advanced research office is interested in better automatic object recognition algorithms for imagery gathered by drones.
What can you see? (Mapbox / Flickr)

For intelligence agencies, unmanned aerial vehicles (colloquially, drones) present a lot of promise for overheard surveillance.

The UAVs can fly around, even over difficult terrain, and capture extensive images of what’s going on below. But then comes the challenging part — parsing through all this data.

This is where the Intelligence Advanced Research Projects Activity (IARPA), a group within the Office of the Director of National Intelligence that deals in “high-risk/high-payoff research programs,” is looking for help. In a challenge posted to, IARPA offers prizes totaling $75,000 for algorithms that can enhance and restore images “acquired under less than ideal circumstances.”

IARPA is looking for two things: image enhancements that can aid analysts looking through the images by hand and algorithms to improve automatic object recognition. The latter is important because of the sheer volume of data the drones can collect.


“Human analysts cannot manually sift through data of this scale for actionable intelligence information,” the challenge statement reads. “Ideally, a computer vision system would be able to identify objects, events, and human identities of interest to analysts, surfacing valuable data out of a massive pool of largely uninteresting or irrelevant images.”

The statement goes on to note that machine learning will be helpful with this task but that there aren’t any off-the-shelf components the agency can use because “they do not take into account artifacts unique to the operation of the sensor and optics platform on a small UAV.”

The challenge is being organized and evaluated by the University of Notre Dame — winners will be announced in May 2018.

Latest Podcasts