A virtual roundtable on analytics

CYBERSPACE—There continue to be more questions than answers regarding video analytics. It has still yet to be determined how this technology will be best brought to market, where it will be best employed, how it will be best made profitable. At ISC West and in the weeks following, this reporter had the opportunity to ask seven of the leading companies the same basic questions, so I thought I'd share with you some of their answers.

Participating in this virtual roundtable are:

David McGuinness, CEO of ObjectVideo;

Scott Schnell, CEO of VideoIQ;

Zeev Farkarsh, CEO of ioimage;

Steve Russell, chairman and founder of 3VR;

Craig Chambers, CEO of Cernium;

Elan Moriah, Americas president for Verint;

and Eric Eaton, CTO at BRS Labs.

These guys didn't really know they were participating in a virtual roundtable, but they were all asked the same two questions and these are their answers, some of them edited for length. Also, some of them said the same basic thing, so not everyone answers both questions.

SSN: Where are you seeing actual installations and traction with video analytics?

Farkarsh: Most of our business is coming from the mid-range market. I can't say the mass-market is really perfect for analytics yet; it's still not quite the mass market. We didn't achieve the hype yet. We're still on the way to the hype. We thought we'd get there 2009, 2010, but the slowdown might change that. But there's still the argument that we can save them some money.

I'm a strong believer in the mass market. I think the big projects are driven by the mass market, not the opposite. It reminds me of the strategy of Microsoft, not going to the giant companies like IBM, but going to the mass market, and they proved that was a good strategy. And I think the same thing will happen here as well. It's nice to have analytics that's affordable.

That was our pitch for the last three years, that was our number one pitch: You're retrofitting the dumb security installation and just adding a box. You're upgrading your security for a reasonable price. We've been doing that for the last three years, and now those competitors are doing that as well. And that's great—people are accepting that's what's got to be done. I believe the new installations will be smaller because of the slowdown, but retrofitting is something you can do, and if someone can offer you an affordable solution they'll go for it.

System integrators need a reason to go back to their customers.

Russell: It's tipping. It's always been ease of deployment and prices that has slowed the adoption of analytics, and we've solved those problems. We have a new product designed for exactly that middle market. We're at the forefront of the industry, we've always had the best analytics, etc., but we were a high-end system. So what's changed recently is that we've been very aggressive in making analytics easy to deploy by our partners, and we're releasing products that are designed to compete head to head with your standard hybrid NVR and DVR out there. From a cost standpoint, we're a no-brainer. It's really just a question of 'Do you want a dumb DVR or do you want to pay the same price for a smart, searchable 3vr?' We make that price not an issue.

Chambers: Where it is being used successfully, and how can integrators use Perceptrak, is when an organization might have one or two, maybe three, people who have the ability to do communications, dispatch, coordination, that type of thing, and they just use Perceptrak as a tool to bring their attention to things as they're happening in real time so they can act on them.

In a campus setting, the Johns Hopkins installation for example, Catholic University in DC, University of Minnesota, every one of those has seen the ability to upgrade real-time security with guards who can intervene as things are happening. That's a huge paradigm shift.

Where that fits with the integrator is that they'll talk with the security operator, ask things like, 'What kinds of issues are you dealing with?' And then they'll recommend a suite of capabilities that, if it includes video surveillance, will often employ Perceptrak or another real-time notifier. This provides the ability to upgrade technology and products, but also to get professional service fees out of upgrading and enhancing the system for different applications as the needs change.

Schnell: The key thing for me has been the continued growth of business with notable end user organizations, both in the breadth of categories and weight of the institutions, including DHS, the oil and gas industry, some state and municipal customers. And we've had continued strength in the schools. We've seen good strength in both the k-12 side and the university side, and that breadth to me is a great sign. Some of the common elements are that those sectors are healthier, more so than retail and the consumer-focused financial services sector. We've had good success in the markets that are healthy and continuing to invest in security infrastructure.

We've seen a lot of activity by the remote guarding partners, too. There's really two center points: One is what I would call guard augmentation, as in, 'I want to have a certain number of uniformed officers at the site, and want to use a professional monitored service to cut down on that number and to give them universal visibility to the site—analytics-monitored cameras do a better job than human beings in terms of persistent vigilance. Guards will be distracted by one incident, where analytics don't get distracted. That's the high-end phenomenon.

The second market, which is much larger, is a market that hasn't been served by guards because they're too expensive or it's too large of an area for it to be practical. This includes mid-market outdoor commercial, equipment dealerships, rental facilities, perimeters at refineries and other industrial installations. Also things like municipalities, citywide surveillance, something that police forces can't really do. They can't monitor cameras or alerts. We're seeing municipalities team up with private businesses.

Eaton: For us, with the integrators, what they like about our solution is the dramatic reduction in the cost of deployment labor and the cost of ongoing maintenance. Some of the integrators that we've talked to, when they go to deploy, their cost curve on setting up each individual camera and making sure it's tuned has been a real sore point on the deployment cost. Our system that self learns and defines its own rules has been well received.

For one of our customers in a major hotel, we deployed the system and had it up and running and giving them information within three to four days. Of course, the patterns that it's learned are limited to those observations you can make in a three-day period. And it can learn things over weeks as well. It has to have enough observations to build that model. We're talking pretty good information in a few days, and then you can tune it and refine it, and that might take a few weeks.

Usually in four to five weeks, customers find they get high-quality information and it's helping them to be aware of what they're having at their facility in a new and unique way.

And it's not just large customers. We do have customers that start out with four cameras and do get value out of the system. It is actually more expensive to do four cameras, but we might eat that cost because we think it's going to lead to further deployment. We do have customers that do four and eight cameras deployments and are very pleased. There are a lot of trends in the industry that will push analytics technology into smaller and smaller deployments for certain niche areas.

McGuinness: Our customers are manufacturers, and we have them around the world in all regions, and based upon the shipment reports, we're encouraged that it is being spec'd, it is being sold, and it is being deployed around the world. Security and safety applications still drive the marketplaces—airports, rail, maritime, oil and chemical plants—but we've also been very excited about performance in business intelligence, retail, casinos, entertainment and banking. It's security at some points and it's collecting data at other points.

SSN: How will analytics eventually be sold? As a software package, as part of a bigger product, as part of an all-in-one solution?

Moriah: Analytics started as a market, transitioned into a niche, and today it has become a feature. That's part of the industry maturing. Once you've got analytics, it doesn't have any value if it's not on the network and connected to other types of data, and you can make it actionable. By realizing that, analytics will never be a market again, it will be a feature of an overall system. It's like buying a car, what a neat feature it is to be able to plug in your iPod. But you don't buy the car for the feature. You can go to the after market for whatever feature you want. The key decision is what car to buy. That's the backbone. If you have the good platform, if it's built in something that's relatively easy to integrate with, then you have a good backbone. That's where I think it's heading and that's where the buying decisions will be based. The smarter enterprise platform companies will make sure they can integrate with third parties.

McGuinness: The way we go to market is that we believe that analytics is not a standalone proprietary system that's running on servers. It can run on servers, but ultimately we feel that analytics is an intelligent ingredient that makes the overall ecosystem more intelligent and responsive. So that's what the ObjectVideo OnBoard platform provides to the now over 40 manufacturer partners around the world. We're that intelligent ingredient that's in other people's products.

Schnell: The short answer is that at the right time, and if the partnership is correct, of course we'll OEM our technology for third parties. We looked very hard at OV; they're the most notable and well-known company in the space. And I think that the problem that people have in realizing the value that OV brings is that the people who embed OV's technologies have only a fraction of the technical wherewithal that OV has, relative to the understanding of the technology and it's my personal opinion that OV's analytics technology is too complex to be effectively represented and supported by their OEM community, and that's why they don't sell that much. I think that it seems logical to me that the most complicated aspect of a product—the camera's not complicated, everyone knows how to adjust white balance and gain—the challenging part is that if you have a technology that requires calibration and tuning and fiddling, if the OEM of OV sells a tiny fraction of their total volume with OV attached, how good are they going to be at installing and configuring that product? And how good are they going to be at supporting, training and communicating with the reseller network?

I just don't think that the technologies that have been OEM'd are really OEMable. OEM technologies have been turnkey if they've been successful, and analytics are not turnkey yet. They have to be a trouble-free installation. In other vendors' cases, they're just not there. The other issue with an OEM strategy is that in order to be financially viable you have to charge a fortune for the product. The single greatest complaint I get about OV is their pricing strategy. Every single option that you would buy, every single behavior pack, costs potentially hundreds of dollars per camera, and you end up spending more for the analytics than for the storage and camera, which just means you'll only every use it in narrow niche applications.

I think the guys in the OEM business are stuck. The products are hard to install, the channel doesn't understand how to install it, and they're outrageously expensive.

Eaton: There's a reason why everyone's trying all these different approaches, and they've worked to varying degrees in other industries. From my level, if you look at specifically analytics, and what we're doing as an industry to try to bring analytics to a reasonable price point and total cost of ownership, there's still a great deal of innovation happening in that space—changes and refining. It's not a completed problem, where you can put a check in the box and say, “that's solved.”

Given it's an area of high innovation, it makes sense to keep that portion of your technology deployed appropriately. It makes a lot of sense to keep those algorithms in a software basis. And then as certain algorithms become best of breed, it makes sense to then put them into hardware optimization. Then if you plug it into an overall mosaic of how you secure a facility, analytics needs to plug into that as easily and quickly as possible. And some integrators have done some great things to make that happen, and we're modeling analytics that analyzes video and alert information and clips of things people should review coming out of the back side, and given that will be the play for the next several years, interoperability and open standards will be the best way to get the technology to market.

I think you're probably gonna see us stay at a software layer for a lot of the analytics until certain of those algorithms become so well used and perfected that it makes sense to put them into hardware.

McGuinness: We also recognize that interoperability and ease of use is a key factor in adoption of analytics around the world, and that's where OV Ready came in. It's an intelligent analytics network protocol that allows devices and applications on the back ends of VMS systems or PSIM systems to work seamlessly together. We recognized there was a need to open that up as an industry standard.

Chambers: Cernium doesn't sell video analytics, Cernium sells products that use analytics as the differentiating technology that makes them work. Analytics as a standalone won't be successful; you need functions around it that people can actually identify and use. You'll note that with [new consumer-oriented product] Archerfish, there is no mention of video analytics. If you bury down deeply, there might be something in the underlying technology, but that's not what we're selling. We're selling keeping track of things in your life using technology.

Like Exit Sentry. Where there are 46 airports using Exit Sentry, if you asked them how they like their video analytics system, they would look at you like you were crazy. We don't sell analytics. We sell a passenger flow monitoring system. And Perceptrak just happens to use analytics to manage large arrays of cameras without people staring at them all the time.