Nexum Automatics

Industrial machine vision

Machine vision systems for quality control and automated visual inspection: defect detection, measurement and robot guidance. For factories in Madrid and across Spain.

What is industrial machine vision?

Industrial machine vision inspects 100% of your parts with cameras, optics and lighting: it detects defects, measures and reads codes without stopping production. When the defect is hard — textures, welds, variable surfaces — deep learning models classify what a classic algorithm cannot distinguish, completing the automated quality control of your line.

A defect that reaches your customer costs far more than catching it in line. At Nexum Automatics we integrate camera, optics, lighting and software with your line and your PLC. And if the system needs to guide a robot or feed results into your SCADA, we leave it connected: vision, robotics and supervision from a single integrator. Want to understand how it works first? Read our guide to machine vision for quality control.

What we inspect with machine vision

  • •

    How are defects detected automatically?

    A camera with the right lighting inspects every part on the line — scratches, burrs, stains, cracks — and rejects defective ones instantly. 100% inspection, not a sample.

  • •

    Can parts be measured without contact?

    Yes: vision metrology measures dimensions, diameters and positions with tenth- or hundredth-of-a-millimetre precision, at line speed and without touching the part.

  • •

    How is an assembly or package verified?

    The system checks presence and absence of components, caps, seals, labels, fill level or complete blisters, and blocks incomplete product before it gets packed.

  • •

    What is vision-guided robotics?

    The camera locates the part — on a conveyor, tray or container — and sends its position to the robot to pick it precisely. It is the basis of flexible pick & place and bin picking.

  • •

    Can codes be read for full traceability?

    We read DataMatrix, barcodes and OCR of batches and expiry dates in line, verify print quality and log every part for complete traceability.

Technology we integrate

CAM

2D and 3D cameras

Area-scan, line-scan and 3D profile, GigE Vision, from the market's leading manufacturers.

LUZ

Optics and lighting

Backlight, dome, coaxial or laser: 80% of a good vision system is decided here.

DL

Deep learning

Models trained on your parts for defects a classic algorithm cannot distinguish.

PLC

In-line integration

Synchronised with PLC, rejectors, robots and SCADA: the system works at the pace of your production.

Typical industry results in machine vision projects

100 %Parts inspected, versus manual sampling

−80 %Fewer defects reaching the end customer

6-18Typical payback period in months

24/7Constant inspection, without fatigue or breaks

Indicative ranges based on typical projects in the sector; we will replace them progressively with data from our own cases.

How we work

🔍

Analysis

Free feasibility study with real samples of good and bad parts: no proven feasibility, no project.

📷

Design

We choose camera, optics and lighting, and define the accept/reject criteria with you.

🔧

Implementation

We build, program and integrate with your PLC and your line, validated on real production.

🛠️

Support

Training, tolerance tuning and model retraining when parts or formats change.

How many defective parts slip through every month?

Send us samples and we run a free feasibility study for factories in Madrid and across Spain: we tell you whether machine vision solves your case and what it would cost.

Request feasibility study

Frequently asked questions about industrial machine vision

How much does a machine vision system cost?
A simple inspection station — one camera, a presence check or a measurement — typically costs between €8,000 and €15,000; a complete station with custom lighting, rejection and PLC integration, between €15,000 and €40,000; and multi-camera or deep-learning systems, above €40,000 and up to around €100,000. These are indicative ranges for the Spanish market: the real price depends on the part, the line speed and the defects to detect. The feasibility study is free and gives you a fixed quote.
How long does it take to implement a vision system?
Between 4 and 10 weeks from order, depending on complexity: the feasibility study with your samples takes days, the design and build of the station is the long part, and in-line integration is planned within a short production stop. Deep-learning systems add 1-2 weeks of model training and validation with real parts before it starts making decisions. You get a fixed schedule in the quote, with the integration stop agreed with you in advance.
Does machine vision work for my part or my sector?
Almost certainly yes: metal parts, plastics, food, pharma, packaging, textile or wood are all inspected; what decides feasibility is not the sector but the contrast of the defect and how consistently the part is presented. That is why we always ask for good and bad samples before quoting: if the defect cannot be seen repeatably with our cameras and lighting, we tell you so and there is no project.
When is deep learning needed and when is classic vision enough?
Classic vision is enough when the defect is geometric and stable: measurements, presence or absence, codes. Deep learning pays off when there is natural variability — textures, welds, organic surfaces, cosmetic defects — where hand-coding rules is unworkable. We train the model on your parts and validate it against a labelled batch, and you choose the balance between false rejects and escaped defects. In practice, many projects combine both: classic rules for measurement and a model for surface finish.
Does it work at the speed of my line?
Yes: systems are sized for the real pace of your production, from parts per minute to hundreds per second on fast lines. The key is choosing the right camera, lighting and processing, and synchronising the trigger with the part flow via the PLC or encoders. If your line speeds up in the future, the system is recalibrated without changing the mechanics. In the feasibility study we measure real cycle times and confirm the margin before quoting.
What do I need to get started?
Just samples: a handful of good parts and every defective one you can gather, with the defects that hurt you most. With that we run the free feasibility study at our facilities — testing cameras and lighting — and send you a report with real images of your parts, the proposed detection criteria and the quote. No need to stop production or prepare anything at the plant. The more defective parts you send us, the more reliable the study's verdict.
What maintenance does a vision system need?
Little, and predictable: periodic cleaning of optics and lighting, calibration checks whenever the station is handled, and tuning or retraining when parts, formats or material suppliers change. We offer support with scheduled reviews and remote assistance; the system stores images of every reject, so any drift is detected and corrected with data, not guesswork. Reviews are scheduled around your maintenance stops so they never steal a minute of production.
Which company integrates machine vision systems in Spain?
Nexum Automatics is an industrial machine vision integrator based in Madrid with coverage across all of Spain: we handle the feasibility study, the choice of camera and lighting, the inspection software — classic or deep learning — and the integration with your PLC, robots and SCADA. We work for manufacturers in automotive, food, pharma and logistics, on new lines as well as adding inspection to existing ones. The first step is always the same: you send us samples and we tell you whether your case is feasible.

Related services

🏭Production Lines🤖Robotic Installations🖥️SCADA Systems