A face in the crowd: Inside the world of fast-track data processing

A face in the crowd: Inside the world of fast-track data processing

How tracking software can process massive amounts of data and put it at your fingertips in a matter of seconds

Imagine processing massive amounts of video data from surveillance cameras and rapidly churning out usable and reliable information on pick-pockets, drunken hooligans, criminals or even terrorists.

Can authorities track a suspect individual across their journey among a myriad of other people, processing complex data in real time so they know where someone is, or at least has been?

Picture this: The London Underground (the ‘Tube’) carries 4 million passengers a day. These passengers can arrive at any one of the network’s 270 stations, change to any of the 11 lines and other stations at will, and leave by another station (or even the one they entered the network by) with complete autonomy and impunity.

Monitoring this movement of millions are 12,000 surveillance cameras, each potentially sending data every second of every day. Manually tracking just one individual under these conditions – allowing for varying lighting, crowded platforms and all based around a few pixels on a video image – is a lengthy and frustrating experience.

Tracking several individuals through different journeys just exponentially compounds the issues. Where to start, where to continue looking, and all that assumes you know who you are looking for.

This is what Michael Haddy, CEO of Adelaide-based software engineering company Innovation Science, took on with his Rapid Passenger Tracking software.

The patented capabilities of the tracking software can automatically process masses of raw video data, locate the relevant footage and provide a short list of suspects within seconds of reporting an incident.

“The data reduction step just takes a couple of seconds so it’s very quick to get to the set of surveillance video and data out of the network that then requires analysis,” Haddy says.

“We very quickly identify groups of people within our suspect list and then provide those to the operator, and that would be reducing a million people in the network down to just a handful of groups.”

Ticket fraud on the Tube is extremely low, so the vast majority of people, even those planning a crime, carry a legitimate ticket. Tickets passing through entry and exit turnstiles represent a largely untapped security infrastructure.

Although they can be supplemented by other technologies to counter the less scrupulous traveller, ticket transactions provide an ideal indication of when and where the vast majority of individuals enter or exit the rail network.

Ticket transactions therefore contribute basic intelligence that can support more complex analysis tasks. The tracking software pinpoints where and when suspects have been throughout their journeys, from entering the network to arriving at the crime location to exiting the network.

It can also determine if different suspects could have feasibly been in contact with each other during their journey and where and when that could have taken place.

Haddy points out using facial recognition technology alone to filter through that amount of passengers would not be practical, especially when aiming to capture information on suspects quickly.

“The problem is when cameras have low light or are positioned in lowlight conditions and it’s in a crowded environment it’s pretty impossible. Even if facial recognition is possible in those sorts of conditions, the amount of computing power required to track millions of passengers would be prohibitive.

“The Rapid Passenger Tracking solution basically uses facial recognition – whether that’s by a human, or a computer, or a combination of the two – only in the very last step in our processing chain after we’ve reduced the problem to a manageable data set.”

Using the 2005 London bombings as an example of how crucial quick detection is in subway crimes, Haddy says it took more than a week for investigators to manually locate video showing the key suspects at King’s Cross station which subsequently proved critical to the investigation. The search of video data at King’s Cross was a result of “somebody’s hunch”.

Based on various scenario tests in simulation environments created for the London Underground and other networks around the world, he believes the tracking software would be able to isolate video of the bombing suspects within minutes and have allowed investigators to examine complete journeys for suspects the same day.

In fact, it was the 2005 bombings that originally inspired Haddy to develop his system. He says he was sitting on a plane flying between London and New York when, instead of reading the usual airport novel, he began tapping out elaborate graph theory algorithms on his laptop.

Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.
Show Comments