Collaboration across sectors is essential to create the most secure and forward-thinking solutions to effectively meet customer needs.
Recent years have witnessed a substantial impact from artificial intelligence (AI), cloud computing and edge technologies, and their rapid evolution is increasing in speed.
As these technologies progress, collaboration across sectors in the security industry is essential to create the most secure and forward-thinking solutions to effectively meet customer needs.
Experts in cloud and AI-driven technologies from the security industry provided their insights on 2023 technical trends in a panel discussion:
- Mishit Patel, CTO, Genea, cloud-based access control
- Jeff Kunzelman, CTO, Arcules, cloud-based video management
- Zvika Ashani, CTO, Irisity, advanced video analytics
- Johan Paulsson, CTO, Axis Communications, camera-to-cloud technologies
Impact of AI and Cloud Innovations on Video Security
Tim Palmquist: In 2023, what advancements, innovations or trends in AI and the cloud have impacted video management, access control, and data analytics most significantly?
Kunzelman: This last year we have seen major advancements in AI and cloud computing that are transforming video management systems. With the open-source availability of advanced object detection models like YOLOv7 and Detectron2, visual analysis has become increasingly commoditized.
This has significantly improved the object recognition, anomaly detection, and behavior analysis abilities of VMS, making analytics more accessible and prevalent without the need for large AI teams to build and tune models.
Additionally, the rapid uptake of large language models like ChatGPT will open the door to much more intuitive user experiences and richer insights.
Ashani: The biggest advancements in AI innovation are specific, custom end-customer solutions. In the past, video analytics was based on a narrowly trained data set and the solution would often fail to meet the customers’ requirements. This is where personalized AI solutions fill the feature/function gaps.
AI-based machine vision use-cases need to be tailor made to the end-customer requirements, including camera infrastructure, and detection and real-time alerting requirements. The only way to meet these requirements is to provide a personalized AI/machine vision solution.
This holds true for on-premises, hybrid, and cloud deployments.
Patel: In 2023, the field of AI has seen transformative advancements, particularly in Generative AI (Large Language Models) and Computer Vision. This rapid evolution is enabling a new generation of intelligent security systems.
With the latest developments in deep learning algorithms, computer vision systems can now identify objects in videos with a high degree of accuracy and speed. These capabilities are increasingly being leveraged in various sectors, including autonomous vehicles, precision agriculture, and smart security systems.
With the help of AI, we can now eliminate false and nuisance alarms by up to 99%, thereby greatly improving the operational efficiency of security operations.
Paulsson: The proliferation of deep learning at the edge is accelerating. Virtually any new network camera being launched today will have deep learning capabilities that vastly improve accuracy of analytics/AI.
Efficient encoding and cybersecurity are other examples of advanced edge device technologies. These capabilities are the foundation for building scalable cloud solutions as they remove heavy requirements on bandwidth, reduce processing in the cloud, and make the system reliable.
Cloud’s Impact on VMS and Access Control
Palmquist: Historically, VMS and access control systems have operated primarily on local servers and infrastructures. In your opinion, how has the onset of cloud technology either challenged or complemented these traditional setups?
Kunzelman: Cloud is just an implementation detail, but it’s an important one because it represents elasticity in compute, storage, and geographical distributions. Incorporating cloud technology into video security systems has been a journey filled with both obstacles and game-changing benefits.
Cloud solutions allow for seamless scalability as an organization expands. With AI-driven analytics built into cloud platforms, users are empowered to sift through data en-mass more effectively than ever, translating it into actionable security or business intelligence insights.
When blended with cloud compute and cloud storage, this facilitates efficient use of hardware and bandwidth allowing the blended profile to match the needs of the customer today as well as allowing it to easily transform to meet the customer’s needs tomorrow.
Patel: I agree, the growth of cloud computing has transformed many industries, offering increased scalability, flexibility, and often cost benefits. As a result, even industries like physical access control, which might seem removed from the digital world, are being reimagined with cloud technologies.
And yes, historically, everything from card-based entry systems to file access systems were designed for local operations. These systems primarily depended on local servers. However, with the advancement in cloud computing and Internet-of-Things (IoT), businesses are transitioning to cloud platforms like AWS, Google Cloud, and Microsoft Azure.
Ultimately, the cloud is transforming legacy paradigms by leveraging its adaptability, automation, and analytical capabilities to converge physical and IT security.
Paulsson: Cloud technologies create opportunities for better hybrid solution architectures — those employing the advantages of on-premise, cloud, and edge technologies. Functionalities are deployed where it is most efficient, utilizing the best of each instance in a system, adding an increased level of flexibility. The VMS is not dead — it’s more alive than ever, it is just transforming.
Benefits of Cloud Installation for Enterprise Customers
Palmquist: Jeff, what’s your response to the thought that much of the industry thinks the cloud is only relevant for small-scale VMS installations and doesn’t work for enterprise-level customers?
Kunzelman: Frankly, it’s misguided. Contrary to popular belief, the cloud VMS has tremendous value for large-scale deployments; especially large, geographically distributed deployments. Certainly, there are challenges to consider — bandwidth requirements and initial transition costs, to name a couple.
However, with the proper infrastructure and a well-planned transition strategy, enterprises can benefit immensely.
Another benefit is consolidated data storage — structured, unstructured, edge, and cloud — enabling more holistic analytics and operational optimization. So, while the industry may be slow or cautious in adopting cloud technology at the enterprise level, this caution should not be mistaken for an indication that the cloud is irrelevant for large operations.
Cloud connectivity is critical for large enterprises to stay competitive. Rather than dismiss it, our industry must reevaluate the merits of thoughtfully leveraging the cloud.
Moving from Blind Recording to AI Event Monitoring
Palmquist: For many integrators, the promise of analytics has never turned into practical solutions and money. Your thoughts, Zvika?
Ashani: Integrators who only power-on the infrastructure are never going to realize the benefits of an AI-based video infrastructure because 99% of security solutions delivered over the past 20 years blindly record video. They provide no intelligence or operational benefits. Therefore, only companies that are looking to solve an operational problem with machine learning will benefit.
As an industry we don’t need more recording, we need AI engines that can recognize and capture the best video clips or images that trigger operational processes after detection. More video doesn’t help solve problems — it creates operational, maintenance, and liability problems.
The integration of cloud deployments and AI/machine vision technology opens doors to lower infrastructure hardware and power consumption costs. AI/machine vision deployed on edge with cloud infrastructure enables end customers to share expensive compute resources across many video cameras and deployment sites.
Debunking the Myth of Cloud-Based Access Control
Palmquist: With concerns about employees being locked out of a facility if cloud services fail, there’s a prevailing notion that access control systems shouldn’t be cloud-based. Mishit, how do you address these concerns?
Patel: There has been a misconception that losing connectivity to the cloud equates to being locked out of a facility when using cloud-based physical access control systems. This is a myth. Today’s cloud-based access control systems are thoughtfully designed to ensure that employees can unlock doors and maintain access to their spaces in the face of cloud service disruptions or internet connectivity issues.
Cloud system incorporate access control hardware panels that function as independent local units, verifying identities and authorizing door access locally. These systems are even equipped to handle emergency lockdowns without needing a cloud connection.
Technologies That Will Redefine Our Future
Palmquist: Over the last three years, what new technology, capability or market shift has had the most significant impact on the industry? Additionally, what technology will be most prominent in the coming three years?
Kunzelman: Hands down, the rapid commoditization of AI. This commoditization will dramatically accelerate the creation of smart analytics, decision-making processes, and overall workflow efficiencies. We’ve moved from having video management systems that primarily served as passive surveillance tools to the possibility of developing intelligent systems capable of supporting real-time analytics and delivering truly proactive threat detection.
With AI and generative models, we can extract insights like never before. The cloud provided the scalability and data management needed to fully leverage these technologies. While AI will continue driving the next three years, it won’t be in isolation. Blended edge-cloud solutions will grow in importance, optimizing cost, bandwidth, and value.
Patel: Over the last three years, cloud computing and the rapid advancement of IoT devices have driven more companies to approach their security more holistically. The success of managed software as a service (SaaS) platform combined with over-the-air updates for IoT devices, has profoundly influenced the industry, fostering a convergence of IT and physical security.
Historically, there were reservations about transitioning to the cloud. However, those objections have notably diminished. A few years ago, the idea of large corporations or government agencies moving their surveillance to the cloud was almost inconceivable.
Looking ahead to the next three years, the unification of security systems is on an upward trajectory, primarily as customers aim to streamline costs while enhancing their workflows. Recent developments in near-field communication (NFC) — Apple Wallet and Google Wallet — suggest that mobile wallet credentials will gain significant traction in the industry.
But two major pivotal advancements poised to redefine the security industry are AI automations and sophisticated edge computing. While AI cannot replace human verification, these tools are invaluable when it comes to helping security teams know where and when to focus their efforts. Real-time AI predictions will grow in importance by providing key insights to security teams.
Ashani: Video data management will create a market shift. Today, the security industry is focused on video management systems. However, when you add AI to a VMS you are missing the meta-data management function. AI/machine vision use-cases create significant data, and the traditional VMS isn’t built to handle this.
Therefore, any AI/machine vision use-case needs to determine how the video meta-data will be stored, managed, and presented to the human operators. The “human-in-loop” is the most important driver in the next three years. The industry mistakenly thinks that AI/machine vision will replace the human. It doesn’t. The most successful use-cases will augment the human-in-loop.
Paulsson: A lot of things have happened within many different areas over the last three years, but AI advancement is the most significant. In the next three years I believe it will be more of a mix between further AI advancement as well as hybrid/cloud-based solutions.
We’ll also see an increased focus on sustainability, cybersecurity, and we’ll feel the impact of increased regulations. All these issues are, to a greater or lesser extent, connected to each other.
Tim Palmquist is vice president, Americas, for Milestone Systems.
If you enjoyed this article and want to receive more valuable industry content like this, click here to sign up for our FREE digital newsletters!
