Technical guides for machine vision inspection projects
Engineering guides for quality managers and automation engineers who are specifying an industrial camera inspection station for a South African production line.
Before you specify a vision inspection system, four questions decide most of the outcome: can the defect be seen on your own samples, how will the station exchange signals with your PLC and reject, does the check suit rule-based tools or deep learning, and what drives the cost. These guides answer each one, in the order most projects meet them. They are free to read and need no sign-up.
Which guide answers which question
Find the question closest to yours. Each link opens a page that answers it in full, including where the method stops.
Swipe the table sideways to see the guide link and what you get →
| Your question | Guide or tool | What you get |
|---|---|---|
| Can a camera see my defect, and which samples should I send? | Preparing samples | Which good, defective and borderline samples to send, how to label them, and what the written feasibility conclusion contains. Printable checklist. |
| How will the station talk to my PLC and reject the right product? | PLC & reject integration | Trigger, result handover, tracking to the reject, reject confirmation and fault behaviour, with a timing diagram and interface checklist. |
| Should the check use fixed rules or deep learning? | Rule-based vs deep learning | Where each approach fits, what samples each needs, how each is validated and maintained, and a decision table. |
| Why is my station rejecting good products? | Reducing false rejects | A symptom-to-cause table, and how to set, prove and maintain limits without letting more defects through. |
| Which light will make the defect show up? | Machine vision lighting | Backlight, dome, dark-field, coaxial, bar and line lights, with a decision table for shiny, clear, curved and textured parts. |
| What will a station cost, and what should a quotation include? | Cost & budget guide | Indicative rand ranges by scope, component prices, the drivers behind the price and what a quotation should state. A firm quotation follows the sample test. |
| What causes product recalls, and which can our line catch? | Product recalls in South Africa | How recalls work under the CPA and R638, examples from published notices, and a table of which causes a packing line can catch. |
| Why does a barcode scan on the line but fail at the retailer? | Barcode verification and grading | The grading standards, GS1 minimum grades, GS1 South Africa test reports, and what an in-line reader can and cannot prove. |
| Which camera and lens does my check need? | Camera and lens selection | How field of view, defect size, speed and the product decide the camera, lens and interface, with starting points per application. |
| How do we choose a supplier for the project? | Choosing an integrator | Three kinds of supplier, ten questions to ask, contract points and warning signs. |
| How is a vision station commissioned, step by step? | Commissioning procedure | Requirement, device selection, camera and lighting set-up, algorithm building, PLC and reject tests, FAT, SAT and handover. |
| Do you have a template for label checks or line clearance? | Inspection checklists | Six printable templates and a PDF, from label and date code checks to a vision challenge test record. |
| Which of our manual checks should a camera take over? | Manual or automated camera inspection | What people do well, what the research says about manual inspection, and what to automate first. |
| What do quality failures cost us today? | Cost of poor quality | The internal and external failure costs to count, where the figures come from, and a spreadsheet layout. |
| Will the station pay for itself? | Payback (ROI) calculator | A payback and simple ROI estimate from your own figures, with every step shown. A scenario, not a promise of savings. |
| What happens between the first enquiry and handover? | How we deliver a project | Each stage from requirement to sample test, factory and site acceptance, and training, and the written record that closes it. |
| Has this approach worked on a real line? | Project cases | Anonymised project records: the brief, the conditions, the solution and how acceptance was tested. |
| Which station would suit our line, and what would it cost? | Line solutions by industry | Complete stations for 26 lines with hardware, installation and indicative rand budgets. |
| I am new to vision inspection. Where do I start? | Machine vision basics | Six short chapters, from what a station is made of to how it is accepted. |
| What exactly is included in a turnkey station? | What a turnkey station includes | What is in scope, what is not, and which trades you or your line builder supply. |
| Can we keep an image of every unit for complaints? | Image archive for traceability | What to store, retention, and how images answer complaints and recalls. |
| Can the station feed our OEE or MES system? | Vision data to OEE and MES | The data a station produces and the interfaces it can use. |
| Which barcode goes on the unit, the case and the pallet? | Barcode basics | EAN-13, ITF-14, GS1-128, GS1 DataMatrix and QR codes on packaging. |
| What do ISO/IEC 15416, 15415 and 29158 measure? | Barcode quality standards | The grading standards and the GS1 requirements, explained. |
A sensible order for a new project
Most inspection projects move through the same decisions. Reading the guides in this order matches the way a project is delivered.
Prove the defect is visible
Everything starts with your samples. If lighting and optics cannot make a fault stand out from the good product, no software will catch it reliably. The sample guide explains what to send and how to label it.
Plan the line interface
A correct decision is only useful if the right product is rejected and the reject is confirmed. The integration guide covers trigger, tracking, confirmation and fault behaviour.
Choose the inspection method
Measurable, well-defined faults usually suit rule-based tools; faults with variable appearance may need deep learning. The comparison guide sets out the trade-offs, including samples and retraining.
Understand the budget
Cost follows from the number of views, line speed, optics, environment and reject mechanism. The cost guide explains the drivers and what a quotation should state.
All guides by group
Each guide explains how the work is done, where it stops and what you need to prepare. None of them states fixed accuracies or speeds: those depend on your product and line, and are confirmed on your samples. The cost guide gives indicative rand ranges.
Line solutions
Line solutions by industry
Complete inspection stations for 26 South African lines, each with HIKROBOT hardware, installation and an indicative rand budget for a typical application.
Inspection selector
Pick your industry and the fault you need to catch, and find the line solution, the inspection application and the device class.
Costs & business case
Cost & budget guide
What determines the cost of an inspection station: cameras and views, line speed, optics, environment and the reject mechanism, and what a quotation should include.
Payback (ROI) calculator
Estimate payback and simple ROI from your own figures, with a conservative case and every step of the working shown.
Cost of poor quality (COPQ)
What defective product costs a packaging line, the COPQ formula, a spreadsheet layout in rand, and what an inspection station changes.
Planning a project
Choosing a machine vision platform
Compare HIKROBOT, Cognex, KEYENCE and Basler by task, samples, software and complete station scope.
Machine vision basics (series)
Six short chapters: what inspection is, what a station is made of, how it decides, where it rejects, what a camera cannot see, and how it is accepted.
Choosing a machine vision integrator
The three kinds of supplier in South Africa, ten questions to ask, what to put in the contract, and warning signs.
Manual or automated camera inspection
What people do well, where manual checks struggle, which checks to automate first, and how people and cameras work together.
How to prepare samples for a feasibility test
Which good and defective samples to send, how to label borderline cases, what photos and line data help, and what the feasibility conclusion contains. Includes a printable checklist.
Inspection checklists and templates
Six printable templates: label and date code checks, line clearance, packaging and seals, vision station checks, challenge tests, dimensional report.
How we deliver a project
Six stages from requirement and sample test to FAT, commissioning, SAT and handover, including retrofits on running lines.
What a turnkey station includes
What is in a station and what is not, and which trades the customer or line builder supplies.
Engineering guides
Choosing the camera and lens
Field of view and resolution, area vs line scan, global shutter, GigE or USB3, C-mount, telecentric lenses and 3D sensors.
Machine vision lighting techniques by defect
Backlight, dome, dark-field, coaxial, bar and line lights, with a decision table for shiny, clear, curved and textured parts.
Vision system PLC integration and the reject
Trigger, result handover over digital I/O or an industrial network, tracking the failed product to the reject, reject confirmation and fault handling. Includes a timing diagram and interface checklist.
Rule-based inspection or deep learning: how to choose
Where each approach fits, what samples each needs, how each is validated and maintained, and a decision table. Both can run in one VisionMaster solution.
Reducing false rejects and the false rejection rate
Why a station rejects good products, a symptom-to-cause table, and how to set, verify and maintain limits as lighting, optics and materials drift.
Machine vision commissioning procedure
Eight steps from requirement to handover: device and optics selection, camera set-up, algorithm building, PLC and reject tests, FAT and SAT.
Barcode basics for packaging lines (series)
EAN-13, ITF-14, GS1-128, GS1 DataMatrix and QR codes: what each is for and how it is read on the line.
Image archive for traceability
Keeping an image of every unit or batch: what to store, for how long, and how images answer complaints and recalls.
Vision data to OEE and MES
What a station can hand to OEE, andon and MES systems, and over which interfaces.
South African rules
Product recalls in South Africa
How recalls work under the Consumer Protection Act and R638, which recall causes start on the packing line, and which checks catch them before goods ship.
Barcode verification and grading
ISO/IEC 15416, 15415 and 29158, the GS1 minimum grades, why a scanner is not a verifier, and what an in-line check can prove.
Barcode quality standards (series)
ISO/IEC 15416, 15415 and 29158 and the GS1 requirements, with the South African context.
Reference
Insights
Practical articles on inspection problems we hear about from South African lines, written by our technical lead, Frank Guo.
Project cases
Anonymised project records: the brief, the conditions, the solution, and how acceptance was defined and tested.
Questions about these guides
Are the guides free, and do I need to sign up?
Why don’t the guides give accuracies, speeds or prices?
I have a product and a fault but no specification. Where do I start?
Who writes the guides, and are they kept up to date?
Have a specific product and fault in mind?
Describe the product, the faults you need to catch and your line. We will tell you which samples to send and what a sample test would involve.