Staff Spotlight: Deanna Beirne
At ALERT, the work of our faculty, staff, and students is the epicenter of the continuous development of the characterization, detection, mitigation, and response to explosive related threats around the world. In this month’s newsletter, we interviewed ALERT staff member Deanna Beirne, our Senior Director of Research Computing and Technical Program Development to learn more about her work with ALERT.
Deanna, who has worked with ALERT since its inception in 2008, supports the center’s research mission in two key areas. First, her work assures that the multi-university center has a sufficient and robust information technology infrastructure for the work being performed. Secondly, and more recently, Deanna has been providing program management, testbed, and transition support to several of ALERT’s research projects, transition efforts and task orders.
Two of the major research initiatives that Deanna works on are the Correlating Luggage and Specific Passengers (CLASP) and Maturation and Validation of Dielectric Characterization Algorithms DHS Task Orders. CLASP uses advanced video analytics to evaluate data from CCTV security cameras in an effort to develop algorithmic software capable of automatically tracking passengers at airport security checkpoints and associating them with their luggage and items. This automated tracking will be useful to identify certain events within the checkpoint, like theft, or passengers forgetting items at the security checkpoint, while also logging how fast it takes passengers to go through the checkpoint. The Maturation and Validation of Dielectric Characterization Algorithms Task Order involves the development of an advanced passenger screening system used to differentiate between a detected item that could be a threat, like a body-borne explosive, or something benign – like a forgotten wallet. This system could increase security effectiveness and improve the passenger experience by reducing the number of pat-downs needed to resolve body scanner alarms.
Learn more about Deanna’s work with CLASP and Maturation and Validation of Dielectric Characterization Algorithms in the interview below.
As the Program Manager for CLASP, can you describe recent research developments and how this will impact next steps in the project?
Yes, a few months ago the researchers were able to benchmark their algorithms on a new video dataset collected at our Video Analytics Lab at the Kostas Research Institute. Built specifically for this project, the lab is configured with a CCTV camera grid, networked video recording system and a mock airport security checkpoint. The lab configuration leverages actual security checkpoint equipment donated to us by the TSA, Rapiscan and Smiths Detection.
To generate the video data we need for the project, we have volunteers act out scenarios which may occur in real life at an airport security checkpoint, and have the video cameras record them. At out last video data collection event, we had Transportation Security Officers, and even a service dog participate, adding another layer of realism to the video data the researchers have access to.
The performance results of the algorithms developed by the teams at Marquette University, Northeastern University, and Rensselaer Polytechnic Institute were good enough to begin merging the individual software components together, which is the next step in creating a real-time system.
What ongoing challenges have there been with CLASP research?
First, every airport is uniquely configured with completely different layouts and physical constraints. This makes it difficult to find a solution that will work for all checkpoints. Given these limitations, it has been extremely beneficial to speak directly with security stakeholders and visit various airports to get input on a solution that can work across a variety of configurations, while leveraging existing airport infrastructure and remaining cost-effective.
Second, getting video data that can be used to test and improve the algorithms has been a challenge as well. Humans are incredibly good at looking at imagery and interpreting what is present or happening in a photo or video; computer algorithms are getting better, but it takes a lot of work to get them to do what people can do. The algorithms need to be trained to correctly interpret the actions and objects you want the algorithm to understand. This, as anyone can imagine, requires a lot of data that we collect in both lab and live environments. The lab is used to generate specific scenarios that may not occur frequently at live checkpoints. We then need to test and train those same algorithms on video data collected at actual airport checkpoints. Using actual footage from real checkpoints will assure CLASP will be effective in a live airport deployment.
You also provide program management for ALERT’s DHS task order focused on the maturation of dielectric characterization algorithms and a partnership with Astrophysics, Inc. to develop 3D reconstruction algorithms. Can you explain what those projects are and their expected deliverables?
The Dielectric Characterization project is funded by the DHS S&T Screening at Speed program and is led by Prof. Carey Rappaport, ALERT’s Deputy Director. The work focuses on taking data collected by a next-generation Advanced Imaging Technology (AIT) screening system, one similar to what you pass through in the airport security checkpoint, and determining if a detected item is something that might be a threat, like a body-borne explosive, or not, like a forgotten wallet. In order to do this, the team is developing software to evaluate the response of the millimeter waves transmitted by the system when they come into contact with an object on the body. From this response, the software can derive the dielectric constant of the object and determine if it is a potential threat, requiring a pat-down, or not.
ALERT was approached by Astrophysics, Inc. when it began developing its Multi-View Computed Tomography Cargo System. Astrophysics created a scanning system that was groundbreaking in its sensor array and scanning methods, but they were in need of a method to convert the 2D computed tomography (CT) X-ray images of shipping pallets that the system generated into clear 3D reconstructions of the pallets. ALERT had realized prior success in the CT reconstruction domain from the DHS funded CT-EDS Reconstruction Initiative and leveraged ALERT researchers such as Prof. David Castanon(BU), Prof. W. Clem Karl(BU), Prof. David Kaeli(NU), and Fernando Quivera PhD, to work on this new problem presented by Astrophysics, Inc. The work has proven very successful and the developed system is currently deployed at JFK airport.
How do all three of the projects that you describe change the screening process in their respective arenas?
The Dielectric Characterization project will help to enhance the detection capabilities of scanning systems, thereby minimizing the number of pat-downs or secondary inspections that have to be done on passengers. By minimizing the number of pat-downs at a checkpoint, it improves the passenger experience and makes the checkpoint line move faster. The 3D image reconstruction developed for the Astrophysics MVTC system makes imaging of cargo clearer and faster – it helps automate a process which is very manual right now. The CLASP program will help to bring in-line, risk-based screening to the airport security checkpoint. Our goal in all of these projects is to make security screening faster, smarter, and less intrusive – both for passengers and DHS components, like TSA and CBP, who rely on them.
When asked what is the most interesting part of her role, “being a part of research projects that are working on ground-breaking methods for advancing security technologies,” is a highlight for her. As a program manager, her work ensures that ALERT research is transferred into real-world solutions.
“The problems we work on as a center are complex and continue to evolve. Figuring out solutions for those problems, such as – are we able to tell if an object on a person is a threat without a pat-down; can we develop 3D image reconstructions that are detailed enough that a pallet of cargo doesn’t have to be unpacked to be inspected; can we leverage video technologies to make the airport security line faster and more efficient – are challenging, and that makes them exciting to work on” she said.
In addition, to her work with ALERT, Deanna volunteers for Boston-area art organizations like the Boston Center for the Arts and enjoys hiking with her border collie.