The US Department of Homeland Security has teamed up with Google and its crowdsourcing site, Kaggle, to search for new algorithms to identify concealed objects detected by airport security body scanners.
The agency has stumped up $1.5m for a Kaggle competition, which aims to uncover an algorithm that can automate the pat-down by Transportation Security Administration (TSA) officers when sensors detect a potential threat.
The TSA is providing researchers with a dataset of images, some of which may contain “sensitive content”, collected from its latest generation of scanners.
“Participants are challenged to identify the presence of simulated threats under a variety of object types, clothing types, and body types,” it notes.
Researchers are likely to be exploring how deep neural networks can help with image and object recognition from the TSA’s dataset.
Kaggle ran a similar competition earlier this year where various teams trained deep neural networks to identify lung cancer from CT scans.
The DHS isn’t actually supplying images of the public who’ve passed through its scanners. The more than 1,000 3D body scans are images of TSA workers who volunteered to assist the contest, according to The New York Times. The workers repeatedly passed through test scanners at a lab in New Jersey.
Given the risk that people with malicious intent could trick an image-recognition system, the competition is aiming to build technology that complements rather than replaces human screeners, Kaggle co-founder Anthony Goldbloom told the paper.
The TSA is trialing the Kaggle competition as an alternative to its usual procurement process, which is currently locked into algorithms from scanning-equipment manufacturers. As it notes, these are proprietary, expensive and not updated frequently enough.
The US Department of Defense’s similarly has tested alternative procurement methods via its own bug-bounty initiatives.
More on Google and AI