For three days in April 2013 Boston residents were paralysed with fear as they waited for authorities to release CCTV footage of the people who planted deadly bombs near the finish line of the city’s world-famous marathon.
It took that long to publish photographs of the alleged terrorists, brothers Dzhokhar and Tamerlan Tsarnaev, despite more than 1000 law-enforcement officers working around the clock.
Why? Because those officers had to manually comb through countless hours of CCTV footage and other video evidence submitted by the public for images of suspects who matched the eye-witness reports.
The Boston Marathon bombings provided a sobering example of just how challenging the task is to search for persons of interest in crowded public places, not just because of the three-day lag but because Tamerlan Tsarnaev was already in an FBI database of potential terrorists.
“Had Boston’s CCTV cameras been equipped with future technology to search for people of interest, Tsarnaev could have been identified far earlier,” said Clinton Fookes, a Professor of Vision and Signal Processing in QUT’s Science and Engineering Faculty.
“While the CCTV system was absolutely crucial for post-analysis it played no part in detecting the disaster. Also, the sheer amount of video evidence provided by these systems simply meant that timely analysis and a quick response was extremely difficult.
“We are unfortunately still limited by the speed at which humans can process the video.
“Authorities across the globe spend literally billions of dollars each year on video surveillance systems but these cameras record only passively – there’s no widespread deployment of advanced real-time video analytics operating across any city.
“Our detectives need better tools to help them datamine video information, or at the least, to help them narrow their search.”
Professor Fookes is on a mission to automate video surveillance technology so law enforcement and emergency workers can do their job more effectively and efficiently.
He’s been awarded a prestigious Fulbright Senior Scholarship to investigate New York’s large-scale video surveillance systems.
Professor Fookes will work in the Media Lab of the City College of New York on one of the crucial missing capabilities of current state-of-the-art systems – the ability to search for people of interest.
The visit will also enable him to assess the impacts, drivers and impediments of real-time video analytics in one of the busiest and most dynamic cities in the world.
“My project will explore the practical and policy reasons for the lack of adoption of large-scale video analytic systems for monitoring our cities, in order to inform future research endeavours,” Professor Fookes said.
“I will also be working with some of the leading policing and defence organisations – such as the FBI and NYPD – that deploy this technology.”
Professor Fookes has previously developed video analytic technologies for airports as part of QUT’s multi-faceted Airports of the Future research project, and he’s passionate about expanding those capabilities into advanced, city-wide surveillance systems.
“The research is not about giving Big Brother more power but, rather, keeping the public safe, saving countless hours in the tedious analysis of video evidence, and increasing the chances of positive outcomes of critical events due to more timely responses.
“Imagine you’ve lost your daughter while shopping in the city centre,” he said.
“Now, imagine a system smart enough to search every CCTV camera in the CBD simultaneously and in real time for images of anyone matching your child’s precise description – 120 centimetres tall; shoulder-length, blond hair in a ponytail tied with a pink ribbon; short-sleeved, white, knee-length dress with pink belt; pink shoes; Peppa Pig back pack.
“When she’s spotted, the system continues to track her movements while alerting the police about her exact location.
“That’s the sort of functionality we can expect from future city-wide surveillance systems.
“They can alert authorities to crimes as they happen, house fires before the neighbours can react or car crashes before the traffic piles up.
“The potential benefits to the public are enormous. The other advantage is that it will be intelligent machines performing many of these duties, removing the need for humans and the potential for abuse and misuse which we have repeatedly seen. Machines won’t bring existing prejudices, biases or preconceptions. They just learn to recognise patterns of interest.”
Advanced, city-wide capabilities of this nature are still years away but it’s hoped Professor Fookes’ Fulbright project will accelerate the development and help create new strategies to counteract the negatives of our current systems.
His four-month mission begins this October.