Advice to Integrators

It is important for integrators to be educated about the products available and the ever-changing technology. Aaron Saks of Hanwha Vision America recommends integrators choose a VMS that fully leverages the AI features they want to use and then become true experts in that system. “Set up a demo environment. Test it by walking through your space wearing a yellow shirt, then see how easily the system can find that footage,” he says. “If you’re using cameras with specialized analytics or functions, verify compatibility with the VMS. Make sure you can search for features like license plates or Re-ID subjects. Know how to archive and protect video for court use.”

Saks adds that it is imperative that integrators don’t just check boxes on spec sheets. “Ensure you understand how to apply these features so you can deliver real value to your clients,” he says.

Steve Burdet of Axis agrees, adding that many VMS’s now include powerful forensic tools as standard features. “Capabilities that used to cost thousands of dollars are often built in at no extra charge,” he says. “Leveraging these advancements can add significant value to your solutions and help clients get more from their investments.”

Forest Liu of ADI | Snap One recommends integrators be acquainted with regulations and local property laws — more so than even the distributors and manufacturers. They should also know what the data retention requirements are. “They need to be thinking about that,” he says. “For example, in the banking industry, they have requirements for the number of days organizations must retain footage.”

Even the basic questions integrators should be asking are changing, says Darron Parker of IDIS. “Before, it was, ‘What are the spaces you’re trying to cover? What’s the most critical?’” he says. “And then you build a coverage plan with CCTV, and you identify where there are end zones, and life goes on.”

With analytics, however, Parker says the questions are now: What sort of events do you want video of? What sort of things would you like to know about when they’re happening, or even be alerted to the potential of their happening? “One of the questions that we like to ask our end user community is: 'What sort of things are occupying your time right now at work?'” he says. “It’s very broad, but that question can lead to some amazing answers because, remember, cameras are just a set of eyes — analytics make the eyes smarter. So how you interact and engage with the end user — that’s changing, and it’s changing rapidly.” Parker believes the industry as a whole isn’t yet prepared for these changes, partly because many are not asking the right questions.

Liu sees the solution coming in through a partnership of sorts between manufacturers, integrators, and end users. “They’re coming to us saying, ‘We need X, we need Y, we’re looking for X, we’re looking for Y,’” he says. “And hearing what our customers want and need helps us, as both manufacturer and distributor in this industry, to create the solutions and products that will help with these installs and projects.”

Kastle’s Nik Gagvani echoes Saks’ warnings. “There is a lot of hype out there,” he says, explaining that a great deal of the value that integrators provide their customers is sorting through that hype and being, in a sense, their consultants and advisors. “There are a lot of spec sheets that claim a whole bunch of things; my advice would be test it. Test it in real security scenarios, not on the bench, and don’t use vendor-provided videos as evidence of something working or solving the problem. Actually put it in in a live environment at your facility and try it out right before you install it at the customer’s site.”

Something From Nothing: The Myth of Video Enhancement in the Movies

Hollywood has a reputation for playing fast and loose with technology, presenting as reality things that astute viewers might point out as unrealistic or “movie magic.” We asked professionals in the video surveillance field for their evaluations about portrayals of forensic searches as presented on TV and in the movies.

“It’s interesting when you’re watching a movie and you see somebody say, ‘Enhance that image,’ and then they do a reflection off a mirror, and then a reflection off an eyeball iris, and all kinds of wild things,” says Richard Solovyev, product marketing manager, ADI | Snap One. Realistically, he says, most cameras installed residentially or commercially are probably going to be 4K or sub 4K. “So, to be able to go off a reflection of a reflection, that realistically doesn’t happen.”

In some movies, Solovyev says, the surveillance AI seems almost psychic. “However, AI is only as good as the training data that it’s been given,” he says. “It can’t foresee the future. You still need that human element of a person analyzing the data, viewing it and understanding the story of what’s going on in that scene.”

Hans Kahler, chief operating officer, Eagle Eye Networks, Austin, Texas, agrees with the need to take precautions when it comes to AI. “There is a lot of investment and development happening in the rapidly evolving AI space, but we’re really still in the early stages — even the AI that has nothing to do with video surveillance still has a long way to go,” he says. “With that rapid evolution comes new benefits and potential risks. I believe AI should be used as a helper, but it shouldn’t completely replace human decisions or actions.”

Aaron Saks of Hanwha Vision America says Hollywood is way ahead of reality. “Hollywood loves the ‘enhance’ feature, zooming in endlessly and sharpening images beyond realistic limits,” he says. “In the real world, you need a certain number of pixels to identify people, read license plates, and gather details. You can’t magically extract clarity from a blurry image. It’s mostly Hollywood magic.”

Saks says that, although Hollywood movies show every camera doing everything, all the time, “In reality, while we can provide Re-ID, AI forensic search, attribute filtering, and license plate recognition, these features are available only on cameras that support them — and only if the entire VMS supports it end-to-end.”

Most organizations don’t replace all their cameras at once with the latest AI models due to cost constraints and lifecycle considerations. Typically, Saks says, these organizations prioritize. “Choke point cameras may have Re-ID, indoor cameras may use AI, etc.,” he says. “On TV, it looks like you can type in a keyword and instantly search every camera. That may just not be practical for many systems today.”

Steve Burdet of Axis Communications says, “Real-life video forensics is more advanced than ever, but it’s still not as effortless as what you see on TV. Finding a ‘needle in a haystack’ remains a challenge — though modern technology can shrink the haystack or enlarge the needle, so to speak. It still takes time, analysis, and often multiple tools working together.”

In many ways, however, the gaps between Hollywood fictional forensic video capabilities and real life are converging, says Matt Lamb of Bosch Video Systems. “What we usually see on TV is either an exaggeration of capability or depicts an archaic, clunky approach requiring manual human inspection. What you often do not see is the procedure and methodical handling of video, once it is identified as pertinent for an investigation, to ensure compliant usage downstream as evidence.”

As we move towards the exaggerated reality that Hollywood often displays, Kahler believes in a future where language learning models (LLMs), a notable forensic search trend, will grow to be extremely reliable tools. “Eventually, a business owner will be able to ask their security system questions like ‘find instances where customers appeared exasperated in checkout lines,’” he says. “LLMs will also be able to provide context and analysis of search results. VMS manufacturers, AI technology providers, camera manufacturers, and others are working on ways to commercialize LLMs. There are many unknowns, but we do know that cloud-based platforms with an open API are set up for success because they can easily integrate LLM capabilities from many sources.”

Burdet adds, “One of the biggest leaps forward is automation. Today’s systems can filter footage by specific criteria — like location, object type, or behavior — so analysts no longer have to manually scrub through hours of video just hoping to spot something. It dramatically speeds up investigations.”

As for the scenes where a sharp-eyed detective spots something in the footage that no one else could? “That’s less likely these days, unless it was so nuanced or minute that the camera couldn’t understand it or detect the object,” Burdet says. “In most cases, technology now does the heavy lifting.”