Following his involvement in the Axis OPEN events, Josh Woodhouse, Lead Analyst & Founder of Novaira Insights, looks at the direction in which security technology is moving – and outlines the huge impact AI will have on tomorrow’s surveillance solutions
What, in your view, will be the most important tech innovations in the next five years for the security sector?
Three broad categories of technology-enabling innovation stand out: cloud, the Internet of Things, and artificial intelligence (AI). These technologies are not necessarily new by themselves, but the breadth of their potential applications means we are now seeing a more rapid acceleration towards them. For instance, the number of professional cloud-connected video surveillance cameras installed in the Americas grew by 86% in 2023. Additionally, shipments of network cameras with embedded AI-capable chipsets grew by 66% in 2023. That’s absolutely the way the market is moving and something we were asked and discussed at the Axis OPEN events.
What makes those three categories so important?
There is a trend towards convergence of security and non-security workloads – and the same is true of individual security technologies. All three key innovations will make an impact together. Humans respond well to visual confirmation of events so, naturally, video devices have become key sensors, and we see a trend towards the Internet of Video Things (IoVT) which includes a new set of IoT sensors which complement video surveillance solutions. IoT technology allows many different sensor and output devices to be networked together; work on standardising the data outputs between these devices is an ongoing trend. Cloud offers the flexibility of distance and on-demand processing. And AI brings new capabilities to the table. United, these technologies enable a range of enhanced possibilities.
How will AI increase its impact in the security sector?
AI is aiding the development and addition of more advanced applications in the field of video analytics. These include functions such as facial recognition and object detection, and extend to the ability to automatically interpret much more subtle video information. These abilities, over time, will become stronger, faster, and able to detect more nuanced behaviour – and generative AI could also be expanded to cover the entirety of security ecosystems.
Does end-to-end AI already help the security space?
We have already seen the impact of AI in many security applications. One example is the automation of alarm event validation. Some end-users have their workflows overrun with alarms, many of them false and unverified. AI technology available today, working from the camera to the client to the cloud, can reduce false positives, which has a huge benefit on operational efficiency.
Where does security AI go next?
AI will continue to increase efficiency by enabling, for example, predictive workflow suggestions presented to the user in their security management client. It will also lead to the ability for users to find desired footage through natural language. Progressively this will become commonplace, because saying, for example, “show me all instances of the red vehicles in our loading dock in the last 24 hours” is far quicker and easier than sifting through hours of video data manually.
Video analytics aren’t new, but end-user adoption has, until recently, been quite low. Has AI changed this?
Yes. Shipments of AI-capable network cameras are increasing – and so are revenues for standalone analytic software licenses, albeit at a slower rate. There are many ways to implement video analytics. It can be embedded directly into cameras, run on network video recorders, be installed as a software layer on a user’s own server, or it can be hosted in the cloud. That flexibility has helped, but it is generally recognised that the addition of AI has improved the accuracy of video analytics, and it’s this which has led to its greater adoption. It is likely that hybrid AI architectures including a combination of edge processing (on-camera or gateway), an on-premise server or NVR, and processing in the cloud will become a popular approach.
What ethical considerations and challenges are arising as AI technologies become more integrated into security technology, and how are these being addressed?
The ethics of AI use in any domain is a challenge. Legislation is doing its best to keep up with advances in technology, but the pace of bureaucracy means it will always lag. This means ethical thinking has become especially important, particularly as businesses rush to implement new technologies which improve the experience for their users. When it comes to security tech this could relate to agreeing appropriate use, understanding and mitigating any biases in the technology’s application or development, and ensuring the AI has been trained ethically.
Does the pace of change in AI technology favour open ecosystems?
Absolutely. In my presentation at the Axis OPEN events, I highlighted the importance of hybrid architectures, and how open systems provide great benefits here, allowing solutions from different vendors to communicate with one another and more effectively share the AI workload. The more advanced the AI model, the more processing power it requires. Organisations which fail to leverage open ecosystems could struggle to remain at the cutting edge of AI technology – and that’s going to be vital in the coming years.
About Novaira Insights
Novaira Insights is a boutique analyst firm and consultancy, specializing in physical security and IoT. Founded in the UK in January 2021 by industry analyst Josh Woodhouse, Novaira Insights provides a range of services, including research, advisory, and thought leadership content creation. Its research approach emphasizes primary research and data-based insights, drawing from an extensive network that includes manufacturers, software vendors, component and technology partners, integrators, distributors, and end-users. www.novairainsights.com
