The Emergence of Analytics
How this security solution is revolutionizing the video surveillance world
- By Robert Muehlbauer
- Apr 01, 2022
Television and movies have always had interesting ideas regarding the capabilities of surveillance cameras. Crime shows like CSI would have viewers believe a still frame can be “enhanced” to near-infinite clarity with just a few keystrokes. Viewers of 24 spent years watching Jack Bauer identify the one terrorist in a crowd of hundreds. And heroes ranging from James Bond to Batman have surveyed banks of wall monitors and instantly spotted something suspicious or out of place—the clue that would lead them to their quarry.
Of course, most viewers are smart enough to understand that movies and shows exaggerate the capabilities of these devices.
Or do they? When 24 first aired, facial recognition technology was a long way from reality—but today, it is gaining greater adoption. Likewise, cameras and image quality continue to improve, making video frames clearer than ever—even without the ability to “enhance.” Best of all, modern analytics are increasingly capable of identifying objects and people with greater precision, allowing security practitioners the ability to reduce false alerts.
What’s more, analytics are helping users optimize resources and improve operational efficiency. In fact, modern technology has caught up, and what seemed imaginary just a short time ago is now very real. In many ways, analytics is no longer emerging—it is here. Modern enterprises should understand how they can put these new resources to use to dramatically improve their security capabilities and more.
Technological Advances Change the Game
Analytics are not new in concept, but their ability to be effectively implemented was severely limited by factors like bandwidth, processing power, camera quality, and storage issues. After all, even the most advanced algorithm in the world won’t produce much insight if it is being fed nothing but low-quality images.
Large amounts of data are needed to effectively train analytics, and storing that data can be expensive. Even organizations who could afford cloud storage solutions often ran afoul of bandwidth limitations as they uploaded massive amounts of video for processing. This combination of factors meant that, while video analytics had obvious potential, the technology needed to make them even more effective just wasn’t quite there yet.
Advances in edge processing power have changed this. Modern chipsets have improved to the point that much of the processing that only recently needed to be done in the cloud can now be done on the edge devices themselves, saving organizations both bandwidth and storage space. Instead of transmitting raw video to the cloud for analysis, cameras can now analyze the video directly and send only the relevant metadata to the cloud, making it easier—and more affordable—to store, categorize, and recall data when needed. These hybrid deployments balance the advantages of both the edge and the cloud—but more importantly, they have helped put modern analytics within reach for organizations of all sizes.
It is also important to remember that cameras and sensors do not function the same way that the human eye does. Instead, they classify things like color on number scales, and different devices may see things slightly differently, resulting in some data variance.
This is also changing as artificial intelligence (AI) becomes more advanced. Machine learning and deep learning models are becoming more common, and Deep Learning Processing Units (DPLUs) are revolutionizing what edge devices are capable of. They enable AI-based analytics to continuously improve their accuracy and performance by revising their datasets as new information becomes available. The dramatic improvement in the quality of both training data and learning models means that not only are analytics becoming more accessible to end users, but they are becoming increasingly capable of self-improvement, compounding their value over time with greater accuracy and new applications.
A Shift in Mentality
While there are plenty of other factors to discuss when it comes to the emergence of analytics, it is important to at least touch on the role that the pandemic played. Over a very short period of time, businesses’ most pressing needs changed dramatically. Suddenly they needed to be able to count the number of people in the store, identify crowds and queues as they formed, enforce social distancing and masking guidelines, and more. Even amid significant advances in processing power and other factors, cameras were often still viewed primarily through a safety-security lens. The pandemic changed this, as they suddenly became an essential part of fulfilling a wide range of new needs. As businesses sought to cope with the changes brought by the pandemic, they discovered analytics offered business insight and the ability to improve operations. Fortunately, the advances that made analytics more accessible came at just the right time.
People counting tools helped enforce occupancy limits. Heat mapping solutions helped businesses keep their employees and customers a safe distance from another. AI-enabled cameras were used to detect and deny entry to maskless customers. Touchless access control solutions helped to avoid surface contact and video intercoms reduced in-person interactions. The pandemic didn’t change these technologies—advancements were already well under way—but it contributed to a significant change in mentality. Manufacturers, systems integrator and users began to identify new and interesting ways to use cameras, sensors and other technologies. This played an important role in helping new analytics tools reach a wider market. For many, this push opened up an entirely new world of analytics solutions—one capable of not just keeping locations safe and secure, but impacting business operations and intelligence as well.
Driving Further Innovation
With more and more organizations availing themselves of the benefits that analytics can offer, the market for those analytics is growing. As a result, a robust application development community has sprouted, as developers look for new ways to apply object and color recognition, motion and pattern detection, and other analytics, to now solve common business cases that span beyond just safety and security. The rise of open-architecture technology has further encouraged the growth of the development community, as device manufacturers increasingly recognize that they cannot do everything themselves. By embracing open-architecture tools and platforms, today’s manufacturers can effectively future-proof their own devices and ensure that integrators and end users have a wealth of options available to them when it comes to meeting specific needs.
Tools like facial recognition can be used in a law enforcement context—but they can also be used in an access control context. Hospitals, for example, can use the technology to ensure that only authorized personnel can access infectious disease wards or other highly sensitive areas. Likewise, government and military facilities can use facial recognition as an extra line or defense to protect secure locations. License plate recognition can be used to identify the perpetrator of a crime, but it can also be used to identify a delivery vehicle and automatically unlock a loading bay gate. In the context of the pandemic, training cameras to verify that customers are masked is important—but in other contexts, a customer intentionally covering their face might be a red flag worthy of issuing a security alert.
These new capabilities challenge what it means to be a “security” tool—cameras can be trained to identify people loitering where they shouldn’t be or recognize aggressive behavior, but they can also be trained to keep an eye on fall risks in hospitals or monitor heat or pressure levels in industrial facilities. These applications can all have a major impact on safety and security that go far beyond the camera’s traditional surveillance uses. And this doesn’t even touch on the growing business intelligence and operations uses for these devices, which can observe shelves and issue out-of-stock alerts, track customer movement and better plan store layouts, help businesses improve the efficiency of their scheduling, and countless other uses. As these applications grow, so too do the opportunities for developers. Integrators and end users will increasingly have an abundance to choose from when it comes to analytics solutions, and manufacturers’ embrace of open-architecture platform has set the table for a strong future in the industry.
The Future of Video Surveillance Is Analytics
A convergence of factors—including the advent of Deep Learning Processing Units, a robust open development community, more intelligent devices trained on better training data, and a new willingness to embrace cameras for purposes outside of traditional security—has led to an analytics renaissance. With smarter surveillance solutions now in the hands of more enterprises than ever, the face of security has changed. No longer is focused solely on preventing theft, tracking suspects, or identifying culprits after the fact. Today’s technology allows businesses to approach safety and security from a wide range of angles, many of which enable security personnel to respond to events in real-time, improving response times and potentially allowing them to address dangerous situations before they can escalate. What’s more, with modern video analytics, security professionals can now provide value beyond safety and security--now offering the ability to deliver business intelligence.
Analytics has always held promise, but recent technological advancements have brought possibilities that once felt like Hollywood magic into the real world. As new and exciting analytics use cases emerge, the video surveillance industry will continue its transformation. Today’s analytics solutions can have a significant impact on any business—in any industry.
This article originally appeared in the April 2022 issue of Security Today.