Machine vision and automated quality control maturity
Your automated quality control, measured line by line and turned into a roadmap.
10 themes, a 5-level scale. And the action that moves each level to the next.
The framework’s 10 themes, already written from L1 to L5. One company, one business unit, or 300 at once.
Machine vision and automated quality control maturity
10 themes, 5-level scale.
Nordhavn Industries
53 / 100
They measure their maturity with Datamensio
An example
This could be your situation.
Take one company as an example: three sites, three spreadsheets, no shared answer.
Nobody can consolidate.
Nordhavn Industries, 2,400 people in Hamburg, Lyon and Porto. A client asks where the group stands. Each site answers in its own spreadsheet, with its own scales.
Three weeks, a single base.
One Machine vision and automated quality control maturity framework (CMMI scale, 5 levels) assessment launched across all three sites at once, from the managers’ interview notes. The framework was already written, its 10 themes and levels L1 to L5 too.
Two costs avoided before being committed.
A score of 53 out of 100, with the gap concentrated on three themes. The AI companion spotted that two actions duplicated those of another audit. The committee report took one sentence to request.
What it saved them
- 3sites measured on the same base, instead of three questionnaires to reconcile
- 2duplicate actions caught before the spend
- 1committee report, with no manual rework
These figures are an example. They could be yours.
The standard imposes processes. Datamensio says where you stand.
01
The framework is already written
Themes, questions and levels L1 to L5, all written. You do not start from an empty spreadsheet.
02
The score lands the same day
Online, by self-assessment link or in interview. Theme by theme, comparable over time.
03
The gap becomes a costed plan
Every step up carries its action. The AI prioritises on expected effect, not on the order of the standard.
04
Progress can be demonstrated
Campaign after campaign, against your target and against your own past. That is what your board asks for.
The maturity scale
One level, the next, and the action that links the two.
It is this mechanism, one level, the level above, and the action that links the two, that turns a finding into a trajectory.
Is the performance of your automated control systems measured on an independent test image set?
- N1
No performance measurement. Confidence rests on line teams’ impressions and the absence of complaints.
- N2
An evaluation was carried out at commissioning, using the images that had served for setup. It has not been repeated since.
- N3
A test set distinct from the setup images exists for the main stations. False rejections and missed defects are measured at set intervals.
- N4
Measurement is systematic across all stations, rerun after every recipe or model change, and results are reviewed with quality.
- N5
Test sets are continuously enriched with disputed production cases, with documented tracking of performance trends by defect family and by line.
Action to move from L2 to L3
For each critical station, build a test image set kept separate from setup images and including known rare defects, then rerun it at the monthly quality review, logging false rejections and missed defects.
« With Datamensio, we meet our objectives far more efficiently. The ERDF inspection services and our supervising ministry particularly appreciated an approach that gives them reliable data. »
Director, CCI 94CCI Île-de-France
« We believe this is the most suitable solution to scale our transformation project and measure impact according to our needs. »
Maja SucekChief Operating Officer, Interreg Danube
Rarely on its own
Frameworks combine. Put several together to cover your business, or have the AI write yours.
Take your first measurement
What this framework covers
Machine vision brings together the devices that inspect a product or process from images: area-scan or line-scan cameras, structured lighting, 3D sensors, thermography, X-ray. Automated quality control adds the decision chain on top: acceptance criteria, thresholds, automatic sorting, rejection, traceability of the verdict, feedback to machine settings. The framework assesses this whole chain, from lens selection through to the statistical use of inspection results by quality and production teams.
In practice, these are projects that progress well as pilots and then stall. The blocking questions are always the same. How are false rejection and missed defect rates measured, and against what reference image set? Who decides to change a threshold, and is that change logged? What happens when the product reference changes, when lighting drifts, when an operator bypasses rejection because it is holding up the line? A vision system that is not governed becomes equipment that nobody questions or steers any more.
One confusion is worth clearing up: deep learning has not replaced classical image processing, it has complemented it. Dimensional measurement, code reading and presence checks remain the domain of deterministic algorithms that are robust and explainable. Learning-based methods answer the question of variable appearance defects, textures, scratches, marks, where no rule can be written down. Two paradigms, two validation approaches, two ways of managing change. Organisations that treat both with the same process run into trouble on one or the other.
The maturity assessment does not ask whether a system exists. It places the level of control and points to the action that moves you up a rung. A site can have cameras on every line and still sit at level two, for want of a managed image database, a qualification protocol or a feedback loop to the process. Another site with two inspection stations can reach level four. That is the information that lets you weigh buying equipment against structuring existing practices.
In Datamensio, the framework is ready to use. You can adapt it to your industrial context: the AI adjusts the themes, questions and levels to your production type, discrete or continuous, or builds a variant from your own documents, inspection specifications, control plans, qualification reports.
Reference standard: Machine vision and automated quality control maturity framework (CMMI scale, 5 levels)
The themes assessed
Inspection strategy and criticality
Mapping of control points, prioritisation of defects by customer and process criticality, trade-off between human, automatic and hybrid control, alignment with the control plan.
Acquisition and physical conditions
Sensor and optics selection, lighting control, mechanical stability, management of drift and contamination, periodic verification of acquisition conditions.
Algorithms and decision models
Deterministic processing and learning-based methods, justification of the choice by use case, explainability of the decision, version control of recipes and models, thresholds and parameters.
Image data and annotation
Building and governance of image databases, representativeness of rare defects, annotation protocol, labelling arbitration, retention and reuse across lines.
Qualification and control performance
Initial qualification protocol, measurement of false rejections and missed defects, independent test sets, comparison with reference human inspection, requalification after change.
Integration with process and information systems
Connection to PLC and MES, verdict linked to serial number or batch, rejection handling, cycle time, station availability, degraded mode behaviour.
Use of results and improvement loop
Statistical analysis of defects by family, feedback to machine settings and suppliers, indicator-driven steering, effective contribution to scrap reduction.
Skills and organisation
Division of roles between automation, quality and production, site teams’ autonomy over parameters, dependency on the integrator, operator training in reading verdicts.
Traceability and evidence
Recording of sorting decisions, archiving of disputed images, log of parameter changes, ability to reconstruct a control after the fact for a customer.
Investment governance and rollout
Return on investment study per station, standardisation of components, replication of a validated solution to other lines, lifecycle and obsolescence management.
A short version of the framework is available for the online self-assessment.
Frequently asked questions
Does this assessment lead to certification?
No, and there is no certification in machine vision. The point is the transformation of your inspection practices: where your lines stand, what level you are aiming for, what actions get you there. The output is a score per theme and a prioritised roadmap.
How does this differ from an installation audit by an integrator?
The integrator assesses an installation they know, often one they delivered, against technical criteria. The maturity assessment covers the whole organisation: inspection strategy, image data governance, qualification, use of results. It compares across sites and over time.
Do you need image processing skills to answer?
The questions are about steering practices, not code or network architectures. A quality or methods manager can answer most of them. Some technical questions can be assigned to the automation manager in collaborative mode.
How long does the assessment take?
The short self-assessment is completed in a single session. The full version, in collaborative mode with quality, methods and automation, spans one to two weeks, with most of the time spent collecting data from the lines.
Can the framework be adapted to our production?
Yes. Themes, questions and levels can be modified. The AI can adjust the framework to your production type, food and beverage, machined parts, electronics, continuous process, or build a variant from your control plans and inspection specifications.
Can several plants be compared against each other?
Yes. The assessment runs at scale across several business units, with a benchmark between sites and against your own past results. A cross-site roadmap consolidates action plans wherever the same gaps recur.
Is the action plan costed?
Each gap identified opens onto one or more services from the service catalogue, with cost, timeline and expected impact on the score. You make trade-offs on comparable items rather than a list of recommendations.
Where is the data hosted?
In France, with OVH, backed up with Scaleway. No transfer outside the European Union. The AI models used can be selected, including from European providers.