Application guide
Start with the defect—not a camera model
A useful packaging-inspection specification states what a good pack looks like, what must fail, how the product reaches the inspection point and what the line must do after a failure. Camera resolution and software are selected from those requirements, not the other way round.
The strongest quotation brief includes real products, good and failed examples, a running-line video, conveyor dimensions and a ranked defect list. It also identifies which requirements are critical, which are desirable and which still need feasibility trials.
1. Define every pack format and normal variation
List the container, closure, label, print, product and secondary-pack variants that will run. Include dimensions, materials, colours, reflective finishes and tolerances. Normal process variation—such as label movement, pouch shape, foam or carton bow—must be represented so it is not confused with a failure.
2. Build a defect library with measurable limits
For each defect, record its location, minimum size, contrast, severity and expected frequency. Provide known failures and borderline samples. Statements such as “detect a missing cap” are measurable; “find all defects” is not. Where a feature is hidden, specify the additional sensor or process control that owns that risk.
3. Record speed, pitch and product presentation
State packs per minute, minimum pitch, acceleration, indexing and planned speed increases. Show conveyor guides, transfers, rotation, vibration and product accumulation. A high-resolution image is only useful when the product remains in the camera's field of view and focal range.
4. Describe the production environment
Identify washdown, dust, condensation, temperature, ambient light, hygiene, hazardous-area and cleaning constraints. Provide the electrical standard, air supply and available space. These conditions affect enclosures, lighting, materials, cooling, guarding and maintenance access.
5. Separate read, measure and verification requirements
State whether the system must detect presence, measure position, recognise text, verify an expected string, decode a barcode or provide a formal code-quality grade. Each is a different task. Define the source of expected recipe, date, artwork or code data and who is authorised to change it.
6. Specify the failed-pack response
Choose reject, alarm or controlled stop according to the product and risk. Define reject confirmation, bin-full monitoring, secure containment, consecutive-failure action and recovery after a fault. Include the distance from camera to reject and the minimum product spacing.
7. Define PLC, HMI and line interfaces
List trigger, encoder, line-ready, permissive, fault and recipe signals. Name the PLC and communications protocols where known. Decide whether the vision HMI or the main line HMI owns recipe selection, user access and production messages.
8. Agree data and image retention
Specify counts, failure reasons, timestamps, batch references and image storage. Define how long data must be retained, who can access it, what happens when storage is full and whether the system must connect to a site network or database.
9. Write the acceptance test before purchase
Agree the sample set, number of repetitions, line speed, allowed false-reject level and response to an uncertain result. FAT and SAT should test the whole chain: presentation, image, decision, product tracking, physical rejection, confirmation, alarms and recovery.
10. Make quotation boundaries comparable
Confirm whether the price includes conveyor modifications, guarding, reject hardware, electrical integration, software, installation, commissioning, training, documentation, trials, travel and production support. List customer-supplied items and assumptions so competing quotations describe the same outcome.
Use this brief for the next step
Send the completed information with representative samples and line video through the project enquiry. For an existing line, also use the retrofit guide and cost guide.