On April 15, two terrorists planted bombs packed with shrapnel, ball bearings and nails at the Boston Marathon, killing three people and injuring hundreds more.
Surveillance cameras captured the Tsarneav brothers drop backpacks two and a half minutes before the bombs inside exploded. Thankfully, the images captured by surveillance cameras allowed police to stop the pair from wreaking more havoc. But the damage was done.
According to the Heritage Foundation, “of the 50 publicly known terrorist plots against the U.S. foiled since 9/11, 42 could be considered homegrown terrorism.” Though authorities have thwarted most attacks, the tragedy of the Boston bombing serves as a harrowing reminder that there is little room for error.
Cyrus Shahabi, professor of computer science and electrical engineering at the USC Viterbi School of Engineering, is creating a powerful technology that can help police prevent future terror attacks and quickly track terrorists before they strike again.
Shahabi, along with dozens of PhD candidates, postdoctoral students and faculty, has developed “iWatch” or “Intelligent Watch,” which extracts video from a city’s surveillance cameras and stores it in a massive database for computer processing. iWatch analyzes the data to detect suspicious behavior, such as someone abandoning a backpack. The system can then immediately flag questionable footage for officers to review.
“iWatch reveals the where, when, what and who of certain data,” said Shahabi, whose favorite television show is Person of Interest, a CBS drama featuring a computer program that analyzes information collected from public surveillance cameras and other sources to identify suspects and threats before a crime occurs.
iWatch uses USC-designed Pan Tilt Zoom (PTZ) surveillance cameras to automatically detect and track every person’s face that appears in a single frame of video. iWatch can then zoom in, producing high-resolution images of faces. This system, which USC Viterbi Professor Gerard Medioni created, can help identify possible suspects to pursue as events unfold.
Since some of Boston’s surveillance cameras are PTZ, “police could have used our iWatch software to easily locate, access and disseminate high-resolution images of suspects in a matter of seconds,” said Medioni, an expert in imaging technology, including surveillance and automated face recognition.
In the event terrorists elude detection and successfully launch an attack, iWatch can help police quickly identify and apprehend suspects after a crime has been committed.
“After the bombings in Boston, the FBI asked everyone at the marathon to upload all of their videos and pictures,” Shahabi said. “That’s terabytes of data. How do you process all of it?”
In Boston, FBI agents sorted through this massive amount of data. Though they identified the suspects relatively quickly, iWatch could likely have hastened the process by extracting the GPS data from all the cameras and phones used to send videos of the event. From there, iWatch can tag each video with a specific time frame and field of view.
“Let’s say you’re interested in a certain area between 2:45 p.m. and 2:50 p.m.,” Shahabi said. “iWatch can immediately show you only the videos that overlap into this area within this time frame. So suddenly, instead of terabytes of data, you only need to deal with megabytes, and it is much easier to manage as a result.”
This technology may have helped FBI agents identify the bombing suspects sooner, perhaps leading to an earlier arrest. Considering that Dzhokhar Tsarneav admitted he and his brother, Tamerlan, planned another attack in the near future, police did not have much time to catch the siblings before they struck again.
iWatch is a $1 million, two-year effort funded by the National Institute of Justice. Though it is still a prototype and has not been applied to the real world, a representative of the USC Department of Public Safety said it might start using iWatch to improve crime prevention and more strategically deploy resources as soon as next year.
“iWatch is a powerful tool because once everything is tagged in time and space, then suddenly everything becomes linked,” Shahabi said. “Once you connect the dots, you can quickly track criminals and protect innocent people.”
More stories about: Research