It doesn’t matter whether you’re a ter­rorist or a two-​​year-​​old who’s been sep­a­rated from her par­ents at the air­port: if you cause a dis­tur­bance in the flow of air­port traffic, you can also cause severe chaos and eco­nomic damage.

Regard­less of your moti­va­tion, moving the wrong way through a secu­rity check­point is treated as a threat by the Trans­porta­tion Secu­rity Admin­is­tra­tion. Breaches of this sort can cause air­ports to tem­porarily shut down, flight delays, and result in the loss of mil­lions of dollars.

You can imagine the cost and the angst that occurs if you don’t deal with this,” said Michael Sile­vitch, director of Northeastern’s Aware­ness and Local­iza­tion of Explosives-​​Related Threats Center, a Depart­ment of Home­land Secu­rity Center of Excel­lence. “So we dealt with it.”

ALERT’s Video Ana­lytics Sur­veil­lance Tran­si­tion, or VAST, team is working to help min­i­mize the damage. In col­lab­o­ra­tion with the Cleve­land Hop­kins Inter­na­tional Air­port, the TSA, and Siemens Cor­po­rate Research the group—which includes researchers at North­eastern, Boston Uni­ver­sity, and Rens­se­laer Poly­technic Institute—aims to help TSA offi­cers make better use of the copious video data at their dis­posal. Their research was recently fea­tured in an article on the web­site fed​scoop​.com, which praised the work for its ability to “solve major secu­rity prob­lems for airports.”

The data we want is usu­ally just a few pixels in a few frames within giga­bytes of data,” explained elec­trical and com­puter engi­neering pro­fessor Octavia Camps, who leads VAST’s North­eastern arm.

Cleve­land Hop­kins Inter­na­tional Air­port has 350 secu­rity cam­eras installed throughout its ter­mi­nals. At a large inter­na­tional air­port such as Boston’s Logan air­port, that number is more like 1,500. Cur­rently, the stan­dard way to inves­ti­gate the footage col­lected on those cam­eras is with the human eye. VAST is automating the process.

In a project called “in-​​the-​​exit,” three sep­a­rate algo­rithms are being opti­mized to detect iso­lated moments in the footage when people are moving against the flow of traffic. This sig­nif­i­cantly reduces the amount of video that TSOs need to review, allowing them to focus more on the type of breach they’re dealing with.

The ques­tion is can we iden­tify breaches of that type, flag the secu­rity officer, and flag the breach before the person actu­ally gets back into the secure area,” said Silevitch.

The pro­gram, which has so far been oper­ating in a research capacity at Cleve­land Hop­kins, has a 99.9 per­cent detec­tion rate with an average of five false alarms per week.

In-​​the-​​exit has been tested since Jan­uary 2013 and will soon be incor­po­rated into everyday oper­a­tions at the Cleve­land air­port. But it is just the first project to be opti­mized with VAST. The team’s next step is to focus on tracking indi­vid­uals as they move throughout the air­port, passing in front of var­ious cam­eras, which each main­tain a dif­ferent field of view. Using fea­tures such as the color and tex­ture of a person’s clothing and the person’s shape and size, these new algo­rithms will be able to auto­mat­i­cally iden­tify the same person at var­ious points throughout the facility.