Join or Sign in

Register for your free membership or if you are already a member,
sign in using your preferred method below.

To check your latest product inquiries, manage newsletter preference, update personal / company profile, or download member-exclusive reports, log in to your account now!
Login asmag.comMember Registration

How intelligent is AI-based video analytics?

How intelligent is AI-based video analytics?
Many solution providers appear to see the term AI as a necessity to market their products. To be fair, video analytic solutions have come a long way from their initial days. But how intelligent is video analytics really?
There is a lot of dialogue surrounding artificial intelligence (AI) in video analytics at the moment. So much so, that many solution providers appear to see the term AI as a necessity to market their products. To be fair, video analytic solutions have come a long way from their initial days. But how intelligent is video analytics really?

In a recent webinar hosted by the research firm Memoori, Carter Maslan, Founder and CEO of Camio, discussed the potential of video analytics to be far more beneficial than traditional rule-based systems. As GPUs become increasingly popular in running AI systems, large manufacturers like NVIDIA and even smaller startups are seeing more investments in this field.

Maslan pointed out that deep learning has been gaining interest in the recent years as they demonstrate several functions that appeal to the customers. For instance, the visual search algorithms allow customers to search for particular objects, animals, etc.

“Its ability to have enough intelligence to deal with all the different orientations and lighting conditions is really the thing that is impressive,” Maslan said, adding that the technology actually picked up in 2012 and 2013. This, coupled with the computing power of GPUs that can take the workload that used to be prohibitively time consuming and resource hungry, has been the fuel that has enabled the wave of innovation.

“Neural networks, generally, not just deep learning which is deep nets that people call convolutional neural networks, are much more capable now,” Maslan said. “There is just great resources available and great tooling. We have been impressed even in the last year how much tools and model developments are out there.”

Companies joining the party

Given the obvious potential of AI, some of the large technology giants have been keen to enter the field and explore opportunities. They are, to a large extent, serving as the trailblazers in this and encouraging the development and adoption of solutions that are more intelligent and intuitive.

“Certainly, the big tech companies like Google with their TensorFlow model has been a great aid,” Maslan said. “Intel has got some great tools that will take the TensorFlow model and import it to exploit whatever resources are on them, the chips that you are running on. So, you have a really flexible mix of CPU and GPU. Resources are available that will target that for you. Then you have all the people that are working on ways to enable and develop training centers that are critically important. So, there is like a whole ecosystem that is forming around the problem of how you get machines to see and understand what is happening.”

There are so many dimensions and layers to this because, fundamentally, it is the assembly and labeling of data. The numerous compilers that can get your models running most efficiently on various chipsets, the data analysis tooling that you would want to have for the output of these things, all contribute to the potential of AI in video analytics.

“When I looked at the physical security building landscape, at trade shows in the past, I was like ‘why is everything so vertically siloed?’, Maslan continued. “You walk the trade floor and you see people selling variations of the same thing from different brands. It doesn’t feel like a technology industry where you have collaborating partners that are exploiting each other’s strengths at different layers.” However, he now feels that such collaboration is happening at some level, with companies offering their strengths to others to join hands and innovate in the field.

IT plays a bigger role

One of the major drivers of AI in the physical security sphere is the integration of the IT sector. The latter is constantly on their toes with vulnerability issues and this would encourage them to unify all the different systems and make them IT-oriented.

“I think there is a combination of factors,” Maslan said. “One, end users that grew up using mobile phones are, all of a sudden, saying, why isn’t there seamless integration as I enter into my office? Then you have IT groups saying, why have these physical security systems been in this isolated area that doesn’t know how to talk modern SSL and REST APIs over regular kind of SaaS-style integrations that they are doing their other businesses, from CRM, ERP, and others. So, I think these two things are creating a sea change.”

In short, the industry is definitely seeing a shift towards going beyond the traditional rule-based analytic systems. As more companies work together, further innovations in this area would be inevitable.
Subscribe to Newsletter
Stay updated with the latest trends and technologies in physical security

Share to: