Transforma Insights recently published a report ‘AI-enabled IoT video analytics: functions and architecture’ which explores one of the fastest-growing intersections of Artificial Intelligence and the Internet of Things: the use of cameras and analytics to generate insights, automate processes, and enhance decision-making across industries. This blog post summarises the key findings from that report, focusing on three main areas: the technology changes enabling the evolution of video analytics, the market opportunities and use cases driving adoption, and the architectural elements that underpin effective solutions.
The deployment of video analytics has accelerated due to simultaneous advancements in AI, computing, and connectivity. Cameras have evolved from passive sensors into intelligent endpoints capable of real-time interpretation, allowing organisations to automate complex visual tasks and integrate them into operational systems.
The growth of computer vision libraries and large-scale AI models has removed many technical barriers to adoption. Open-source tools such as OpenCV and YOLO, combined with large training datasets, now enable accurate object detection, anomaly recognition, and behavioural analysis. These capabilities make video analytics accessible to a much broader audience, reducing reliance on bespoke development. Vision-language models extend functionality further by combining visual and contextual understanding, supporting applications that require situational awareness or natural-language input.
AI-assisted programming interfaces now allow users to describe analytics requirements in everyday language. Instead of writing code, users can issue instructions such as “alert me when a person enters a restricted zone.” This natural-language interaction lowers the technical barrier to deployment, accelerates configuration, and opens the field to non-specialist users.
Connectivity plays a critical role in enabling reliable video transmission and analysis. The rollout of 5G networks, with their high bandwidth, low latency, and features such as network slicing, has allowed real-time analytics in sectors like emergency response and industrial automation. In parallel, edge computing and adaptive streaming techniques optimise performance in constrained environments by processing data locally or reducing video resolution when bandwidth is limited.
Hardware innovation has broadened the scope of viable video analytics deployments. Modern cameras include embedded AI processors and enhanced low-light performance, enabling analytics in challenging environments. While prices have stabilised, functionality continues to improve, making systems more capable at similar costs. Some architectures embed processing directly within cameras, but many rely on edge gateways that handle analytics for multiple devices, improving efficiency and flexibility through containerised workloads.
Beyond individual components, a growing focus is on orchestration platforms that manage the integration of video analytics, AI models, and business logic. These platforms are emerging as a crucial layer in enabling effective deployment and scaling of solutions. AI orchestration includes the management of models, the integration of analytics outputs with business processes, and the triggering of automated actions. More discussion on AIoT platforms can be found in ‘A diverse range of AIoT platforms is emerging, many focusing on distinct aspects of the AIoT device lifecycle’ (July, 2025).
Developers benefit from a mature ecosystem of AI frameworks, cloud-native tools, and containerisation technologies. This environment accelerates experimentation, simplifies deployment, and ensures scalability. Together, these technology shifts have transformed video analytics from a specialist domain into a mainstream capability embedded within connected systems.
Video analytics is extending the boundaries of what connected systems can measure and automate. It delivers richer insights than traditional sensors and has clear benefits across multiple verticals.
Video analytics enhances surveillance by identifying anomalies, unauthorised access, and safety breaches. Automated alerts and behavioural recognition allow faster responses and improved situational awareness for governments, schools, and private organisations.
Beyond loss prevention, video analytics is now a business optimisation tool. Retailers analyse shopper movements and dwell times to refine store layouts, while hotels and restaurants use analytics to monitor occupancy, service delivery, and customer satisfaction. These capabilities link visual insights directly to commercial performance.
Cameras mounted on vehicles or dashboards, combined with AI models, detect fatigue, distraction, and risky driving. In emergency vehicles, real-time video analytics integrated with 5G connectivity provides situational awareness during incidents. This dual focus on safety and efficiency is driving widespread adoption in both private and public fleets.
Factories and construction sites are applying AI-driven video analytics to monitor safety compliance, detect product defects, and analyse workflows. Automated PPE detection and process tracking reduce risk and improve productivity. Integration with enterprise systems ensures insights translate into immediate operational actions.
Airports, highways, and rail networks use analytics for traffic management, crowd control, and automated number-plate recognition. These applications improve safety and streamline passenger or vehicle flow. As urban transport systems become more connected, the demand for real-time visual intelligence is rising sharply.
In offices and retail centres, video analytics supports both security and operational efficiency. AI systems detect unauthorised entry, monitor occupancy, and link data to building management systems for energy optimisation.
Ports, warehouses, and distribution hubs employ analytics to track assets, monitor worker activity, and ensure compliance. Cameras combined with AI models enhance visibility, reduce losses, and improve safety in large-scale logistics operations.
An emerging trend is the use of cameras as intelligent sensors that send metadata or event-based messages instead of full video streams. This “video sensing” approach reduces bandwidth needs and enables real-time automation in constrained environments such as remote industrial sites or areas using satellite connectivity.
The report highlights future potential for user-programmable, plug-and-play devices that combine a camera, a gateway, and computer vision libraries. These easy-install systems allow users to configure functions via natural-language commands, opening the market to a wide range of horizontal applications beyond traditional vertical solutions.
Collectively, these use cases demonstrate how AI-enabled video analytics is becoming a foundational capability across industries, transforming how organisations monitor, understand, and automate their environments.
Video analytics solutions are built on a layered architecture that integrates hardware, software, networks, and orchestration. Each component contributes to scalability, interoperability, and security.
At the top layer, applications translate raw visual data into insights relevant to specific business objectives. Many are built as full-stack offerings combining cameras, analytics, and video management software. Open computer vision libraries and natural-language interfaces have democratised these tools, enabling low-cost but capable deployments. The differentiator increasingly lies in integration and managed services such as onboarding and maintenance.
Analytics platforms manage AI models, data flow, and orchestration between devices and business systems. They perform model management, trigger actions, and optimise connectivity. These platforms bridge the gap between AI innovation and operational environments, allowing modular functions to be deployed through APIs and integrated with enterprise applications such as ERP or CRM systems.
Modern cameras have evolved into intelligent endpoints with onboard processing, enhanced imaging, and open interoperability standards such as ONVIF. Managed services for updates and security ensure reliability. Edge gateways often complement cameras by handling heavier computation while enabling scalability and easier upgrades.
VMS platforms control recording, playback, and device management. Increasingly cloud-based, they incorporate event tagging, metadata generation, and integration with open standards. Their role is central to coordinating video streams with analytics workflows.
Edge gateways process data locally to reduce latency and bandwidth use. They host containerised AI models and business logic for rapid response and data privacy. This architecture supports constrained connectivity scenarios and enables real-time automation.
Connectivity options range from fibre broadband to mobile and satellite. Cellular networks, particularly 5G, offer flexibility and enhanced performance through network slicing and API-based management. Bearer-aware compute adjusts video processing based on network conditions, ensuring continuity and cost efficiency.
Governance is a critical component. Solutions must adhere to security and privacy standards through encryption, access control, and up-to-date software inventories. Oversight frameworks ensure responsible AI operation, transparency, and compliance with regulations across different jurisdictions.
Beyond the technology stack, deployment success depends on service delivery. Installation, model training, and support services are increasingly integral. Providers who combine these with hardware and software into modular packages can accelerate adoption and provide greater assurance to enterprises.
AI-enabled video analytics is reshaping the IoT landscape. Technological convergence has democratised access, allowing organisations of all sizes to derive value from visual data. While analytics capabilities are becoming commoditised, differentiation now lies in orchestration, integration, and compliance. Emerging opportunities such as video sensing and easy-install programmable devices point toward a future in which cameras function as flexible, intelligent sensors across diverse industries.
The report upon which this blog post is based, ‘AI-enabled IoT video analytics: functions and architecture’ is published as part of Transforma Insights’ Advisory Service. Additionally, as a counterpart report, we recently published ‘AI-enabled IoT video analytics: market landscape’ which briefly profiles 142 companies that feature most prominently in the discussion of remote video analytics, particularly in scenarios where the functionality might replace IoT sensors.
We also recently participated in a webinar with Semtech and Digital Barriers looking at the video surveillance market: The End of Video Throttling for Mission-critical Video Surveillance.
