Actionable Intelligence

Turning misconceptions of AI into dynamic security solutions

• By Bill Brennan
• Apr 01, 2021

There is no doubt that Artificial Intelligence (AI) is here to stay, as new AI-driven solutions are pervasive across every category of professional and consumer electronics today. This is especially true in the professional security arena – both in terms of physical and cyber security – as exciting new AI-driven solutions promise to move the industry from passive monitoring to predictive analysis modes of operation. The ability to autonomously collect and analyze vast amounts of data from holistically integrated security systems also opens-up a myriad of new business intelligence applications across the enterprise.

Yet, despite the pervasiveness of AI technology, it is still widely misunderstood, and that is understandable after hearing comments from some of the most respected technology gurus of our time who portend that AI will be death of civilization or believe that sci-fimovies portray an accurate picture of the future. It is important to separate media grandstanding and fiction from the real science behind AI and the pragmatic benefits it promises to deliver. For the purposes of this discussion, let’s focus on the latter purely from the perspective of the professional security market.

Video surveillance, access control and other related physical security systems have progressed in quantum leaps over the past several years. It was not all that long ago that we were all looking at grainy B&W images from analog CCTV cameras, and access control was relegated to keys and guards who manually logged individuals entry/egress activities with a ballpoint pen.

The truth of the matter is that the seemingly rapid evolution from analog to digital to networked, and now intelligent state of technology, has actually been in the works for quite some time – actually since the 1940s. This is when research first began on developing computational models that would eventually evolve into artifi- cial neural networks, which provide the foundation and processing power for AI and machine learning technology as we know it today.

According to Wikipedia, artificial neural networks, otherwise referred to as connectionist systems, are computing systems that are inspired by biological neural networks like those in human brains. Such systems have the capacity to “learn” to perform tasks by considering examples, generally without being programmed with taskspeci fic rules. This is essentially how the term “artificial intelligence” and the alternate term for AI, “machine learning”, came into being as a means of describing a new generation of smart machines and devices capable of making autonomous decisions.

Many of today’s new physical security solutions actually live up to the long-standing expectations of AI with the ability to autonomously learn and analyze situations and then make decisions based on the data. These new solutions employ one or more of three different electronic learning techniques: Supervised learning. Where continuous feedback is presented to formulate decisions.

Unsupervised learning. Where decisions are based on estimation of statistical data received by an input source or device,

Reinforcement learning. Where rules are adjusted based on a combination of estimating data inputs and pre-established rules to reach the desired results.

There are several relatively new, yet highly familiar professional, security solutions trending today that adhere to these AI models, including intelligent AI surveillance cameras, dynamic and adaptive facial recognition, video redaction, and physical and software enabled IoT sensor technologies. When used in combination, the aforementioned solutions transform traditional security monitoring and recording systems into sources of actionable intelligence enabling new forms of predictive analysis.

Unfortunately, it often takes a cataclysmic event to identify the need for new and innovative security solutions that migrate into the mainstream thereafter. The ongoing pandemic is no exception. New integrated AI-driven security and health safety solutions have made their way to market to help address current challenges and compliance mandates.

• Automatically identifying, authenticating and tracking individuals upon entering a facility or campus.
• Automatically identify individuals who are not wearing face masks in environments where masks are required.
• Automatically detecting and identifying individuals with elevated body temperature.
• Automatically restricting access or redirecting individuals who may require additional screening or attention.
• Redacting innocent individuals’ faces from video that is being used to apprehend suspects or as video evidence.
• Detecting anomalies in the environment or standard operations that pose a potential threat or liability.

Although we now perceive these intelligent integrated solutions as specific measures to combat the pandemic, we should anticipate that they will become permanent criteria for safe and secure facilities and healthy buildings moving forward. A perfect example of this was the widespread adoption of metal detectors and body scanners following 9/11.

This level of integrated intelligence represents the tip of the iceberg in relation to the wide range of new and emerging professional security applications for AI-driven software and hardware solutions. This is relevant to new and relatively inexpensive sensors and analytics that can easily supplement traditional surveillance cameras and software solutions such as VMS and access control management platforms.

The additional data from new sensor solutions further increases situational awareness by providing the ability to detect potentially threatening anomalies. Examples include everything from audio analytics to detect and identify screams, gunshots, explosions and glass breaking to environmental sensors that recognize noxious fumes and gasses, smoke and fire, and feel vibrations such as explosions or earthquakes.

As we constantly hear on the news these days, it is important that we “follow the science” as it pertains to AI – which is a win-win for security professionals, users and integrators alike.

This article originally appeared in the April 2021 issue of Security Today.