Smart Manufacturing

Smart manufacturing: why visual intelligence is the factory’s hidden data asset

Phil Black - PII Editor 28 minutes ago

28 5 minutes read

Smart manufacturing: why visual intelligence is the factory’s hidden data asset

Key points

Listen to this article

Manufacturers have spent years connecting machines, systems and workflows. Gabriele Mangiafico, Axis Communications, explains why video now belongs in the smart factory data
layer.

Rising operational costs are a pressure point UK manufacturing cannot ignore. In recent independent research commissioned by Axis, 57% cited costs as the major challenge, a problem that sits alongside quality and visibility as areas where manufacturers most want better operational insight. Factories already collect data from production systems, maintenance platforms and planning tools, but video, one of the richest sources of evidence, is still often left out of the equation.

Cameras already sit across production lines, warehouses, loading bays and quality checkpoints. Yet only 18% of manufacturers use video technology to support productivity, even though 39% see productivity as its greatest future value. To treat cameras as assets reviewed only after incidents or investigations leaves much of that value untapped.

How industrial vision AI adds context to smart factory data

The problem in smart manufacturing is rarely a lack of data to work with. Smart factories are very good at recording events. A line stopped or a defect rate increased, for example. But those signals rarely explain the scene around the event. This is where visual intelligence earns its place. It gives operational data the context, the ‘why’ behind an event.

Only 16% of manufacturers fully integrate security technology data into operational decision-making. Visual intelligence can help bridge the gap between operational data and real-world events, showing what was moving, missing, delayed or deviating from the expected. For manufacturers, the value of video analytics is clearest in three areas when context changes outcomes: quality, flow and safety.

Video can contribute to improve quality assurance, for example. Video analytics complement the inspection process, enabling faster reviews, earlier intervention, and a strong evidence trail. The cost of a quality issue rises as it travels down the line; a missed component or assembly error is far easier to fix at the station than it is to correct once the product has left the factory.

The BMW Group uses Axis network cameras as part of the AIQX platform in its iFACTORY smart manufacturing initiative, supporting AI-driven inspection in vehicle production. Real-time images allow production staff to identify defects as part of the production process, with video analytics generating prompts for potential issues. In one application, for example, the system helps to verify whether the correct rating plate has been applied in the correct position.

Using AI video analytics to improve OEE, throughput and defect detection

More directly, video can be used as a core component of industrial process monitoring. Nestlé uses Axis products across its French production sites, monitoring automated and robotic processes for both defects and optimisation opportunities. Compact cameras monitor coffee jar filling stations, for example, allowing operators to remotely supervise the robots on the line and providing evidential footage for later analysis.

For operations leaders, video is a root-cause tool. When linked with production events, it allows teams to compare and understand the reasons for throughput changes, to see what has worked and what has not, and to make fair assessments about overall equipment effectiveness (OEE). Stoppages and repeated interventions get that all-important ‘why’.

And for those pushing forward the digital transformation agenda, the extra value of linked, explained events may be all the argument that is required for video to be brought into the manufacturing technology stack.

Supporting PPE, near-miss and HSE reporting with AI-powered video monitoring

Video’s place in safety is often underappreciated, but as manufacturers invest further in HSE systems, training and reporting, it is risky to leave safety visibility entirely to human operators.

Even experienced operators cannot see every risk, every time. Near misses, PPE non-compliance, and ergonomic or zoning risks are easily missed by tired eyes, but video analytics both help bring these events to operators’ attention and aid in post-incident investigation.

Steel manufacturer ArcelorMittal Belgium employs over 2,700 Axis network cameras across its seven high-risk production departments, allowing edge-based analytics to aid in production line monitoring and restricted area safety. These cameras, attached to machines and monitoring restricted zones, are not bespoke designs.

The company’s AI functions can run on compatible Axis cameras and, in case of a malfunction or damage, engineers can replace a camera within an hour from on-site stock. By simply utilising video in the right way, HSE leaders get answers to their questions, and stronger evidence for intervention.

Using existing camera infrastructure for video analytics

This does raise a practical question: can manufacturers use the cameras they already have? Sometimes they can. Sometimes they need to upgrade, reposition or supplement them. Analytics performance depends on whether the image is good enough for the job.

Lighting, angle, resolution, frame rate and field of view all affect the result. Determining what works begins with an audit to determine what can already be seen, if the image quality is adequate, and which views support measurable operational problems.

The answers to those questions help shape the architecture. Edge analytics – that is, analytics that run directly on camera hardware – are useful where speed and resilience matter; cloud analytics can support broader analysis and comparison across sites. Many manufacturers will need a mix of both, with local detection at the line and central reporting for the business.

Connecting edge AI, cloud analytics and OT cybersecurity

Integration is just as important. Video intelligence should not become just another isolated dashboard. To support smart manufacturing, analytics outputs need to feed the systems manufacturers already use, from MES and SCADA to ERP, historians, maintenance platforms and BI tools. Standard protocols such as Modbus and OPC UA can help bring camera-generated events into industrial automation and control environments.

Cybersecurity and governance also need to be designed in from the start. Cameras are connected devices, and they often sit close to sensitive operational systems. They must be managed with strict access control, lifecycle management, network segmentation and disciplined firmware updates, around rules concerning video retention and worker privacy.

Building the business case for smarter factory vision

The business case should begin with one operational problem. That might be defect escape cost, downtime per hour, time spent reviewing incidents, manual inspection effort, safety reporting workload or lost output from recurring stoppages. Once the baseline is clear, video intelligence can be judged against measurable improvement in quality, downtime and safety.

Start with a problem that the ops team already cares about. Prove the value in one line, process or site, then build an architecture that can extend across quality, throughput and safety without creating a disconnected technology estate. Factories already have eyes on their most important processes. The task now is to turn that view into usable operational intelligence that delivers clear ROI.

In focus: a spotlight on manufacturing by Axis explores how video data can help manufacturers add context to operations, close the visibility gap, and support smarter, safer and more efficient manufacturing.