“Fixed cameras will account for less than 50% of surveillance footage in five years’ time”: Milestone CTO on ‘gigantic’ data and neural networks

Bjorn Skou Eilertsen, CTO of Milestone Systems, was thinking big in the Security Management Theatre on day three of IFSEC International.

Speaking on the topic of how ‘hardware-accelerated video content analysis and the internet of things will transform surveillance’, he reflected on the changes disrupting the industry now and the paradigm shift still to come.

With 90% of the world’s data created in the last two years, the term ‘big data’ doesn’t even begin to encapsulate the magnitude of the data revolution, he argues. “Is big data even enough now? Gigantic data might be better,” said Eilertsen.

Despite the ubiquity of fixed CCTV cameras, they account for a shrinking share of surveillance footage as “mobiles, body-worn cameras and drones” proliferate. “We believe that in less than five years from now, more than 50% of streams managed by video management systems will not be from fixed cameras,” he predicted.

Aggregation, automation and augmentation

A trinity of aggregation, automation and augmentation will equip the industry to accommodate the burgeoning volume of data, said Eilertsen, who joined Milestone in 2013 having worked for both IBM and Microsoft.

“Aggregation happens all around you,” he explained. “Only a few years ago it would be a fixed camera, fixed sensors, very rule-based. But now there are 285 million surveillance cameras in operation. That’s only a fraction, because everything is being captured on mobile.

“With neural networks we can start predicting behaviour.” Bjorn Skou Eilertsen, CTO, Milestone Systems

“How do we automate these things? This is where our vision of intelligent data plays a role. Deep learning plays a role.

“Augmentation: how do we put these things together?

“So a vast amount of information is being gathered. This is why a lot is going to happen on the service side. People think it will be on the edge, out there on a single device.”

Eilertsen pointed out that Data is already being aggregated from multiple sources in an automated process deployed on assembly lines in manufacturing plants.

“Aggregating forms patterns, but it’s so much information – petabyte after petabyte of video and sensor information. What will we do with it?

“Who is going to look at the patterns and figure out what the intelligence is? That is where the important changes are coming in terms of AI, deep learning and neural networks.”

For simple systems with only a few components, “it’s fairly easy to make rule-based analytics and go with the flow.” However:  “When you start aggregating data so big and complicated that humans simply cannot operate them,” that’s where automation and augmentation come in.

Neural networks

The shackles are now off thanks to quantum leaps in technology.

“This has been difficult to do for a long time because conventional CPUs cannot compute fast enough. That’s changing now” with the introduction of the GPU, said Eilertsen. “The GPU is a multicore computer. It changes the way we can make models, neural networks. It makes a lot of different ways of working the data.”

“The days of having one company try and do everything is over in my opinion.” Bjorn Skou Eilertsen, CTO, Milestone Systems

He refers to a prototype that can show “1,500 surveillance cameras, to full HD quality, continuously recording, including motion detection. For those who can’t do the maths, that’s 45,000 frames a second. It is very, very difficult to do on regular computer hardware.”

He says there is a big shift away from “conventional, rule-based analytics” to systems managed by neural networks. “Neural networks, deep learning algorithms and artificial intelligence are not based on fixed outcomes. The problem about today’s analytics is it’s a predetermined outcome.”

“With neural networks we can start predicting behaviour,” he says.

However, human operators will still have a role to play. “How do we make machine intelligence combine with human intelligence? The point is to enable people to make faster and better decisions.”

He says this new paradigm has huge potential in the field of body-worn video for law enforcement. You can “take all the aggregated media from years back, days back, minutes back, and time-lapse it.

“They identify all different objects and put them into a sequence, so a one-hour video can be reviewed in one minute. That’s a really good example of how we start adding human interaction based on machine learning. It really makes it a lot easier to work with these systems.”

Collaboration

Collaboration with partners has long been part of Milestone’s modus operandi, but its importance is growing further still.

“The aggregation, automation and augmentation will transform the entire industry,” says Eilertsen. “But it’s impossible to do alone.

“For a very long time it’s been everyone on their own trying to make their own analytics a little bit better than the rest. But it’s really holding back innovation. What Milestone and the Milestone community is really about is enabling everyone to participate.”

If he’s correct about the industry’s direction of travel then the changes ahead are nothing short of revolutionary.

“The days of having one company try and do everything is over in my opinion. We all need to collectively move forward.

“I think in five years when we look back at the industry, we’ll have two ways of looking at it.

“One person will say: ‘Why did we miss it, why didn’t we see what was happening? The other, more interesting way is: “How did we use our imagination, how did we change the rules, set the agenda and change the industry?

“We need to think as a community. We need to start innovating together, and we can move a lot faster.”

Free Download: The Video Surveillance Report 2023

Discover the latest developments in the rapidly-evolving video surveillance sector by downloading the 2023 Video Surveillance Report. Over 500 responses to our survey, which come from integrators to consultants and heads of security, inform our analysis of the latest trends including AI, the state of the video surveillance market, uptake of the cloud, and the wider economic and geopolitical events impacting the sector!

Download for FREE to discover top industry insight around the latest innovations in video surveillance systems.

Download The Video Surveillance Report