On July 9, 2018, the Department of Homeland Security (DHS) Science and Technology Directorate (S&T) in conjunction with the U.S. Transportation Security Administration (TSA) announced eight winners of the Passenger Screening Algorithm Challenge. Contestants were challenged to develop automated algorithms that will lower false alarm rates at airport checkpoints. The eight prizes, totaled $1.5 million and ranged from $500,000 to $100,000.
About the Competition
The competition was held in two stages. As part of the first stage, contestants were challenged to develop an algorithm that analyzed 1,000 volunteer passenger images hosted in a variety of different formats. Contestants who made it to the second stage where challenged to replicate the analysis, but with a much larger dataset. The winners of the competition will have their algorithms used alongside existing systems to undergo testing and evaluation. Ultimately, DHS S&T and TSA hope to utilize these algorithms at airport security checkpoints soon in the future.
Why it was Launched
TSA and DHS sponsored the Passenger Screening Algorithm Challenge to identify non-proprietary algorithms that can “improve speed, detection of prohibited items, and accuracy” in both Government-developed and Original Equipment Manufacturer-developed (OEM) screening equipment. The competition is part of the DHS S&T Apex Screening at Speed (SaS) program. Through non-traditional procurement methods, and creative research, DHS and TSA hope to find screening solutions that can be implemented within five years, which will help TSA operate more efficiently and securely, while minimizing inconvenience to passengers. Apex SaS envisions an environment where 300 passengers and their carry-on luggage are scanned per lane per hour. Both DHS and TSA want Agents spending less time reviewing images, and more time observing and assisting passengers.
TSA attributes delays at its airport security checkpoints in part to the performance of its current detection algorithms and their procurement process. TSA currently purchases algorithms exclusively from the manufacturers of its scanning equipment and has complained that these algorithms are expensive and often take a while to be fully released to the agency. Additionally, the proprietary nature of the algorithms means that changing them to fit new screening or threat detection requirements is challenging often must be performed by the OEM on the equipment itself. Contestants are thus challenged to develop open-source solutions to these challenges and help DHS/TSA discover “new, non-traditional performers that might otherwise be overlooked”.
Open Source Code
The U.S. Government has been making more of their source-code public to encourage improvements through collaboration among the public and between government agencies. By making the code open-source, agencies can identify potential reuse of existing code and not have to use funding to develop similar code. As part of this movement, the U.S. Government launched a website that makes available a portion of the U.S. Government’s code. The Federal Government’s open source code policyis also available for anyone to read.
