12.6.24 – SSI – D. Craig MacCormack –
Security integrators can use video analytics and artificial intelligence to help customers make decisions, if they know what they’re doing.
The continuing growth of artificial intelligence (AI) means there is exponentially more information available every day. But, if you don’t know what to do with the reams of information you access every time you generate a report, you won’t be able to use it to make more informed decisions.
Security integrators are at the forefront of the AI revolution, helping their customers understand how to use the video analytics tools that are built into just about every video surveillance camera on the market these days so they can act more quickly and prevent problems before they happen.
Mike Compton, CTO for Everon, has been managing software development for the company since his company was acquired in March. He says the sales team’s requests for AI and video analytics integration have become more popular, ranging from the server side to camera-side edge-based analytics.
“I think it’s getting much more popular, frankly, because I think it’s starting to work better,” he says. “Historically, there’s been a lot of false positives. There’s been a lot of ‘the camera’s got to be in a very perfect spot and the right field of view’ and I feel like the analytics have gotten a little bit better and a little bit more robust.”
“There’s not so much pressure put on the installation side. I think that has lessened a little bit over the last couple of years, so I think that’s built some confidence in the technology. Because of that, now end users are a little bit more interested,” says Compton.
Part of that change, he says, is because we’re past the early adopter stage for AI and video analytics.
“End users now have confidence and then that just runs through the industry, so I think that’s opened the door quite a bit,” says Compton. “I think there’s a lot of interest. I think it really depends on what vertical you’re in as to what you’re actually interested in.”
“There’s a lot of models that are out there that do a lot of different things. Some apply to a specific customer group. Some don’t apply in any way, shape or form. So, I think it really depends on what your environment is, what vertical you’re in,” he says.
Everon is “doing much more integration work around the metadata side of what the models will deliver,” says Compton, and they’re getting requests for different types of video analytics and AI tools.
“Most people are really asking for people detection, object detection, vehicle detection,” he says. “LPR (license plate recognition) is a big one. In the retail space, you still have people that ask about queuing, counting dwell time, those types of things. Now, those things actually work a little bit better than they did historically.”
Artificial intelligence is “becoming a major part, maybe the engine that drives it all” for the video surveillance-related installations for Sellersburg, Ind.-based Security Pros, says director of client technologies Mike Williams.
The company works with global clients and U.S.-based end users, mostly in the Midwest, across several vertical markets, including municipalities, critical infrastructure, trucking, financial services, schools and parks.
Ted Curtin, the founder and president of South Hadley, Mass.-based Repworks, a security and AV manufacturer’s rep firm, says the company has been “pushing analytics for many years, but there are several roadblocks and we’ve kind of been working around those.”
“The first analytics we were exposed to were by Bosch about 15 years ago,” he says. “At the time, we were working with Bosch, and they had a lot of German engineers working on advanced camera features. I had never heard of analytics before.”
“I escorted some of the Bosch engineers around to some high-profile end users in the Boston area and, listening to them, I was shocked to find out that a lot of the end users expected analytics to take over their access control and other parts of their system intrusion and that was beyond anything I had thought of,” says Curtin.
About 5% of the attendees in Repworks’ training classes say they’re using analytics, he says, a number that’s far too low for his liking.
“If you learn analytics, you become a better competitor, because everyone has this same fear [that] they did of networks back in the year 2000,” says Curtin. “The first barrier to analytics becoming day-to-day in most of the systems is not the manufacturers that are producing amazing analytics. Dealers don’t feel comfortable integrating it and promoting it and the end users don’t understand it.”
“The second barrier is that analytics are more useful to the end user beyond the security department. HR has a tremendous opportunity with analytics,” he says. One example is insurance companies requiring those in multi-story dwellings to always use handrails in staircases.
“It’s beyond just the security,” says Curtin. “Marketing has been using analytics for years. Why is Pepsi using blue and Coke using red and which color in the end caps is the most popular? We’ve been pushing analytics to end users for their marketing department for years to see how many people dwell using heat mapping at a certain location.”
Jonathan Kozak, chief technology officer for Elite Interactive Solutions, says the company is “always trying to get out ahead before a crime has been committed, before damage has been done.”
“Hunting for the technologies and bringing them together to be able to do that effectively has been quite a challenge,” he says. “Over the years, the technology has been lagging behind our vision, so it’s very difficult to piece it together to make something that actually works.”
Elite has become “extremely effective at the endeavor of crime prevention and bringing a real solution to our customers rather than just the recorded video that shows their property being damaged or somebody being injured, etc.,” says Kozak.
Everything Elite does has an alert that includes video and images, he says.
“We do not accept simply a line of text or a burglar alarm where it’s just a line of text that arrives to a command center,” says Kozak. “We want to be able to really see and comprehend what’s going on so we can relay that information to the appropriate emergency response authorities that we’re contacting to handle a situation. Video analytics factors into that very heavily as they are going to be the root source for a trigger.”
Other alarms include fire alerts or glass breaks or a panic button, he says. These alarm types don’t come with images or video on their own, says Kozak. Rather, Elite’s technicians configure the systems to automatically bring up images from cameras associated to those alarms’ time and place.
Who’s Leading the Conversation?
Larger enterprises are typically “pretty sharp,” says Compton, making the point that there doesn’t need to be a lot of explanation about the merits of AI and video analytics with those customers.
“They understand the industry,” he says. “They know what options are out there. They go to the trade shows. They see what’s on the floor. They understand the technology that’s coming.”
“When the lines blur between video analytics and you get into some of the AI conversations, I think there’s some confusion there about what you can and can’t do and really decipher what the customer is truly looking to accomplish,” says Compton. “I do think the customer base is much more educated and they’re a lot smarter about what the technology is doing and what’s available to them. They are asking more questions about it from a sales perspective.”
There are still some aspects of AI that’s “not quite ready,” says Compton, “so there’s a lot of setting expectations that I think is important from a sales side back to the end user. But I feel like end users are [far] more educated today than they’ve ever been.”
The glut of information means end users today come into the sales process armed with a surprising amount of knowledge, he says.
“Five, 10 years ago, most of your education was cut sheets and things like that,” says Compton. “Today, you can go self-educate online and everybody’s websites are way better than they used to be and there’s a lot more information that’s readily available.”
Williams says customers are more interested in finding a solution to their problem rather than worrying about what specific technology is being used to solve it.
“If you were a manufacturer of drill bits, you might be under the mistaken impression that your customers want drill bits, but no, they want a hole,” he says. “Your customers want an answer. They want a capability.”
Financial services companies, for example, want to ensure that one person is never alone in certain rooms or areas to protect critical infrastructure and information, Williams explains.
“There are access control ways of doing that and there are video surveillance ways of doing that if you have staff to watch the cameras 24/7, which nobody has anymore,” he says. “How you do that is less important to them than that you do it, and that it works reliably 100% of the time.”
“We had been doing that exclusively with access control and a two-man rule and making sure that you can’t just have a single person read into this room and have the door unlocked. You need two people to read in in order to unlock the door. That’s one way of getting that done,” says Williams.
“I think a better way, and certainly a way we’re going to be proposing to them is, ‘How about we get rid of all that but one camera, have AI watching it and alerting you if ever a single human is in that room for more than a few seconds?’ Obviously, it’s going to be one person at a time coming through the door. So, you need to fine-tune that. But that’s what’s driving all this. ‘Give me an answer that you can put your name on, that will work,’” he says.
Williams says customers don’t give any more pushback to AI- and video analytics-embedded solutions than any others and that the disagreement is usually centered on price.
“If you get a no, that means you haven’t convinced them of the value,” he says.
Integrators, says Curtin, are “always looking for a way to be better bonded” with their customers and “analytics gives an excellent opportunity for the larger integrators to fairly easily integrate analytics.”
Embracing AI and video analytics “comes from a vision of what’s possible and an understanding of how to bring a real solution to the customer,” says Kozak.
“The customer is not asking for AI,” he says. “I suppose a lot of customers do actually ask for that as it’s a buzzword and you can expect that but they’re not asking for that. That isn’t really what they want. What they want is a real solution. And that is what we are offering right now. And that’s what we’re going to expand upon.”
“The true solution that we’re offering to them is the actual prevention of a crime or the actual prevention of a disaster. I think there is a bit of misunderstanding there too. It’s not the analytics and the trigger to the command center that prevents the crime. It’s those items followed up with a very near real-time response,” says Kozak.
Boosting Recurring Revenue with AI and Video Analytics
Compton sees more opportunities for recurring revenue stemming from the continued growth of AI and video analytics across the security sector.
“I certainly think it opens the door to do more,” he says. “I think it makes products more effective than they were historically. If you think of a traditional video system, a lot of it was very, very reactive.”
“With some of the intelligence that’s being pushed to the edge, and to the server side and even some of it that that’s run cloud side, now there’s just that ability, through the analytics, to be a little bit more proactive and do more real-time analysis of what’s happening at that current moment and then alert people to that and I think that opens up a lot of opportunity,” he says.
The real value that security integrators can bring to their customers, says Compton, is “what you do with the information.”
“If you take that information and you go pile it up on a hard drive somewhere and it just sits there, and no one is interacting with it, there’s no alerts or notifications being driven off of it, you’re not using that information to do something differently than you did five years ago, what’s the point?”
“If you’re going to leverage that information to provide more services around it, that’s where it gets interesting to the customer, and that’s where the opportunity to drive RMR and more service-based revenue comes,” says Compton. “I think there’s more things that you can do for the customer to help support them.”
“There’s lots of data, and there’s lots of noise It’s making sense of the data and making sense of the noise that’s the value. The value is not the analytic itself. The value [lies in what] you do with it when it triggers an alert, or triggers a notification, or tells you that something’s going on. Are you actually responding to that? If the answer is no, then I probably wouldn’t put it in and I would save my money,” he says.
“If the answer is yes, you’ve got a solution to take action on that and be actionable with what information is being generated, then there’s a lot of value in that,” says Compton. “There’s more that can be sold and more services that you can package around that.”
Being able to cut down or eliminate false alarms helps alarm companies, helps dispatchers, helps local police departments and helps customers, he says. That’s just one use of video analytics that could make it worth it for end users.
“I think the other area where it gets interesting is the data that gets generated from the analytics and being able to make better business decisions overall,” says Compton. “It’s not just a security solution or a loss-prevention solution.”
“Can we actually provide business metrics back to the organization to where this investment is more interesting for people other than just the sliver of security folks [who] are involved?” Compton asks rhetorically. “Can we bring operations into it, bring marketing into it? Are there other things we can do with the tech stack that actually drive positive behavior or actionable information back to those other departments as well?”
“There’s interesting data that’s being generated,” says Compton. “It’s all about leveraging that to its full capacity.”
AI has moved beyond the bleeding edge, but “it may still be leading edge,” says Williams.
“If you get people who’ve been burned on new stuff before, they’re probably not a candidate for moving into this realm,” he says. “But if it’s somebody who is either so driven by a problem that they have that they have not been able to solve with existing technology or we’ve had a couple of people that that just want to be first.”
RMR “really is the lifeblood of the industry,” says Williams. “If you can know on Jan. 1, based on just the revenue you already have on the books and is almost certain to come in this year, that you’re going to be profitable, that buys you all kinds of business credibility, with your bank, with your employees, with your finance people, with your owner.”
“It’s not just in the alarm industry that people expect to pay a subscription. People are expecting to do that in anything that involves a computer now and that’s because of the advantages of cloud and the biggest one there is nobody wants to be responsible for cybersecurity,” he says.
“These systems need to be maintained,” says Williams. “A big part of that is making sure they’re secure and only the right people can get into them. But a big part of that, too, especially in regard to our industry, is it has to work right. It has to sit there doing basically nothing for months on end, but when the critical event happens, it has to be 100% up. You need ongoing attention to it, and you need to be paid for that. So, that just lends itself to a subscription model.”
Recurring monthly revenue is an important factor for security businesses whose owners are looking to sell their companies, says Curtin.
“In the ’70s, ’80s and ’90s, a lot of integrators started their businesses and grew them and now they’re all at that age where the kids don’t want to be in the business and the value of the resale of their corporations or companies is the recurring revenue,” he says.
“Analytics does offer an opportunity to increase recurring revenue based on the model and that’s a two-pronged sword,” says Curtin. “Some companies base recurring revenue on different charges for different analytics. Other companies are now migrating video management software to a one-time fee. The dealers have the ability to be creative and make it part of their package, so the opportunity for recurring revenue is very real.”
Security integration is a business model that “does revolve around RMR,” says Kozak.
“We find ourselves in competition with integrators who are installing cameras,” he says. “They install cameras, they install an NVR and then they wash their hands of it. That’s the end of it for them, and they kind of hope that’s the end of it except for an add-on.”
“We, on the other hand, stay with the customer indefinitely. We expect this to be a lifelong relationship that we have. So, we have contracts for service and SLAs involved for things that matter the most for us to be able to do our job of preventing crime and catastrophes at their location,” says Kozak. Elite’s technicians take proactive steps to ensure problems are addressed quickly.
Which Verticals Embrace AI and Video Analytics?
Retail is among the vertical markets that have been the most welcoming to using AI and video analytics, says Compton.
“There’s a variety of different things they’re looking for,” he says. “You’ve got the traditional, ‘I want to see how many people are coming in’ and conversion rates, queue metrics and heat mapping. We’re also seeing a lot of requests for, ‘can you help detect aggressive behavior that’s in the parking lot or in the front of my convenience store?’”
“That concept of leveraging analytics to be smarter about what’s happening in your store, especially after hours, where a traditional alarm system would come into play, applies. They’ve all run into the same issues with false alarms and potentially false alarm fines and those types of things,” says Compton.
Law enforcement is “all over” AI and video analytics, says Williams.
“If you have any customers that are in that space, or you’re looking to move into that space, you need to lead with advanced capabilities of all the eyeballs that they already have around the educational market space right now,” he says.
“If you’re not talking about this with your schools, whether it’s K-12 or higher ed, you need to be talking about this right now,” says Williams.
The adoption has been slower in logistics and some of Security Pro’s other vertical markets, he says.
The construction industry is one that’s fully embracing analytics, says Curtin.
“Anytime an inspector comes on, he’ll shut a project down if he finds people without hard hats, without vests on,” he says. “The analytics today can tell you if someone removes a hard hat or vest or walks onto a site without it. It’s easy to demonstrate to them how that works.”
“Thermal cameras offer a better opportunity for outdoor detection with analytics because they’re not trying to process a whole pile of color information. They’re processing basically black and white, so the analytics used in thermal cameras are actually being adapted quicker,” says Curtin.
Repworks has a growing number of customers in the cannabis industry, he says.
“Whenever we do the long-range fence detection systems using thermal cameras, we always add analytics to that, because it instantly picks up anyone trying to cut and cross in, especially during harvesting time to grab mature plants,” says Curtin.
On the other hand, “the average project, which might be a printing press or a small manufacturing company, doesn’t see the value, although it’s there as an example,” he says.
“People counting is another one that’s already well-established,” says Curtin. “The retail industry is already looking at end cap displays and looking at how many people stop at different displays to analyze the effectiveness of the marketing campaigns that’s already occurring. The problem is that a lot of our dealers don’t even mention that.”
“Casinos, as an example, can tell you which machine draws the most time on a slot machine floor at what time of the day and they can actually move the machines around and actually play with the customer attendance based on that. So, they’re adapting analytics quite heavily,” he says.
Car dealers were among the first to embrace AI and video analytics, says Kozak.
“I think that’s actually well-known in the industry now, but I think they had the most obvious need, the most obvious immediate problem for outdoor perimeter protection, and for unwanted visitors overnight on their property,” he says.
“They wanted to have no fences on the properties, no gates to make it look welcoming,” says Kozak. “But then that also welcomed people at night made them feel welcome and invited and so we were able to shoo them away with our two-way audio.”
Among the other vertical markets deploying AI and video analytics are multi-family housing and shopping centers.
“We want to apply it behind a video camera to watch what is happening in the scene and be able to trigger a noteworthy event for someone to handle as an event,” says Kozak. “They both can use a boost. We’ve seen video analytics, the edge analytics, come a very long way.”
“Most video analytics are a bit incompetent to what we are trying to do, which is crime prevention, and that is because they were primarily made to do smarter recording to save storage space on the NVR rather than detect unwanted activity,” he says.
“The dream is to see chips that are so powerful that they can do true AI on the edge running heavy code commoditized so that you can have true behavioral analytics, which is something I do think we are going to see very, very soon,” says Kozak. “We could be having true behavioral analytics in 2025 as services that we can put out.”
True behavioral analytics could mean AI recognizing a pattern of someone walking through a parking lot trying to get into every car and questioning the person about their intentions, he says.
“This is what I think everyone imagines and that they want from analytics,” says Kozak. “They have that sort of expectation. But the reality is that is still just not quite here, and there are huge cost barriers involved, as well, to have an entire server farm for graphical processing units and the power consumption that would take to be able to analyze live video streams in HD and 4K at 30fps.”
The Future is Data-Driven
While it seems like there’s already more information than anyone could possibly use available to them at the click of a button, the advancements in AI and video analytics and the adoption of those tools are still in the early stage, says Compton.
“Interest is very high,” he says. “In most conversations I’m involved with with end customers, the question is asked pretty consistently in terms of full-scale implementation. It’s not yet becoming a standard in every install that is happening.”
“The ability to leverage analytics is being included much more often than maybe it had in the past. I think a lot of edge devices like cameras have the capability. I don’t know that I’m 100% convinced that it’s being implemented all the time and used consistently,” says Compton.
“There’s a lot of conversation around, ‘hey, it’s there and it’s on the camera and when you’re ready to turn it on, we can turn it on and make use of it,’” he says. “I think there’s probably, frankly, a little bit more of that than it’s being consistently leveraged across the masses.”
“It’s probably still in the early stages of implementation but I think it’s moving pretty quick. I still think a lot of it is driven by what the customer wants. I think, as the customers see that this stuff works and they’re more educated on it, and they see how it can bring value to their business, they’re going to ask for it, and that’s going to drive the adoption rates,” says Compton.
Compton expects big advancements in the use of AI and video analytics in the next few years.
“I personally believe that leveraging analytics to be smart about the decisions you’re making around central station monitoring dispatch is going to significantly accelerate over the next five years or something like that,” he says. “I think that’s going to significantly accelerate, because it just makes too much sense not to.”
“If you look at the challenges that are out there, false alarm rates and those types of things, that’s a problem that everybody’s asking to be solved, so I think analytics is a great way to begin solving that problem. I think that will really start to accelerate, because we’re going to see some success in those areas. And I think adoption will be pretty quick when we start to see success in those areas.”
“The other piece of it is how quickly can platforms really leverage data that’s being generated from the environment that’s not a camera? Once that happens, you can make use of the data that’s being generated, and you can drive action on that, and you can do that really effectively,” says Compton.
“There’s a way to go” when it comes to full-scale adoption of AI and video analytics, says Williams.
“We are not in the widely accepted range,” he says. “We’re in the early adopter phase. We’re past the bleeding edge right. We’re to the point where people who have that burning need are happy to let us fill it with this technology.”
“We’re probably not to the part of where it all tips and mass adoption. A camera is just a mute witness. That’s not good enough. I need something watching my cameras all the time and alerting me to things that that it knows I’ve cared about in the past,” says Williams.
“I think we’re well on the way to mass adoption but it’s going to take [some time] to get us to the point where of course, you have AI watching your surveillance system, watching your access control system, looking at the two of them together to make sure that this alert coming from the access control system is real,” he says. “I don’t need a human with the technology available in the market already today. The AI can do it for me.”
Williams sees a bountiful future for AI and video analytics.
“Ten years ago, customers were asking me, ‘Where video surveillance going to be 10 years from now?’ and I got it mostly right,” he says. “I think 10 years from now, certainly not one and probably not three, but 10 for sure, I feel very confident that if you are putting in just a video surveillance system the way we do it today and it really doesn’t do anything other than let you look up things that have already happened, that’s going to go by the wayside.”
“It’ll still happen and some people, that’s all they need, but that’s going to be less and less of the marketplace. More and more people are going to want to know what’s happening in real time and more and more people are going to want to never log into their video management system,” says Williams.
Analytics “are very sophisticated and easy to use now,” says Curtin.
“They’re very, very mature and every month or two, they come out with a new analytic that gets added in for free,” he says. “It might be three years before the end user becomes educated enough to ask for analytics.”
“Slips and falls are one thing right away that you can do. We tell nursing homes that they can have that in their day room. If someone falls down, it’ll give them an analytic. People who run ATMs are already looking at anyone kneeling. If you have an ATM, that’s an analytic that someone’s messing with it, so it’s become very vertically oriented as to the channel,” says Curtin.
“What I’m hoping for is a broader look at analytics and what can it do for me,” he says. “We tell our people when you’re talking to the security director of a corporation, have them introduce you to the facilities manager and the HR people.”
“If we agree that exposure to end users is key to the growth of analytics, the manufacturers could do a better job in marketing to the end users. I think that could be a big boost to the use of AI and analytics,” says Curtin.
AI and video analytics are beyond the early stages, says Kozak.
“It’s matured far too much to be called an early stage for our purposes of crime and catastrophe prevention,” he says. “That’s mostly for edge analytics. The AI is still nowhere near that dream of a true behavioral analytics live in the cloud. That is quite some time out. We’re more likely to see that occur on the edge before we even see that in the cloud, because of chip advancements that are going to be coming out. Some are already out.”
Kozak expects the AI and video analytics sector to become more commoditized and more effective in our daily future.
“The analytics applications running in edge devices are very expansive now, doing license plate recognition, facial recognition, smoke detection, fire detection and intrusion detection,” he says. “They have scheduling; they have their own notifications. They have inputs, outputs, and those are all put to use. And we can expect that to continue.”
About the Author
D. Craig MacCormack, Digital Editor
Craig MacCormack is a veteran journalist who joined Security Sales & Integration in June 2023 as digital editor. He covered AV, IT and security with SSI’s sister publication, Commercial Integrator, from January 2011 to June 2021.