Synthetik Awarded Contract from DHS to Generate Machine Learning Training Data for Passenger Screening
Synthetik Applied Technologies has announced back in August the award of a contract from the U.S. Department of Homeland Security (DHS) Science and Technology directorate Apex Screening at Speed program to use deep learning to provide a next-generation passenger baggage screening capability to TSA. This six-month project is funded under the Small Business Innovative Research (SBIR) program, a program established by the Small Business Administration Office to ensure that the nation’s small, high-tech, innovative businesses are a significant part of the federal government’s research and development efforts.
A record 4.1 billion airline passengers traveled during 2018 – and according to the International Air Transport Association (IATA), this number is on-track to double to 8.2 billion passengers over the next 15-20 years. To meet this challenge, a new wave of passenger, baggage, and vehicle scanning technology is being developed and deployed at airports globally.
However, in order to truly serve the billions of airline passengers traveling annually, automation is urgently needed to improve the speed and effectiveness of security screening, reduce wait times, and increase detection accuracy. A machine learning (ML) based approach for automatic detection is the right choice. However, there is a key issue: to train effective machine learning-models a large volume of high-quality data is essential, and manually generating such imagery is time-intensive, laborious, and expensive.
To meet this challenge, Synthetik is developing a new physics-based synthetic data generation and annotation platform that will provide millions of training examples to support next generation high-accuracy object detection at speed and scale.
According to Peter Vonk, CEO at Synthetik, “…we are very excited about this project as it builds directly from our other ongoing programs with DHS. We’re already working on 2D and 3D passenger baggage screening and vehicle scanning, and this project will help provide the high volume of ground-truth training data we need to launch at global-scale. Synthetic data generation will unlock the potential of machine learning and change security screening forever.”