The cap is the last component to be placed on the container and, at the same time, the one that guarantees everything described above. A defective closure can lead to leaks, microbiological contamination, or loss of carbonation in carbonated beverages. In addition, it is the first part of the product that the consumer touches.
On production lines exceeding 30,000 or 40,000 units per hour, visual inspection by an operator is simply unfeasible. That is why cap inspection using machine vision has become a standard in the packaging industry.
The objective of a cap inspection system is to verify 100% of containers in-line, without stopping production. Its function is threefold: detect deviations, reject non-conforming containers, and record data for traceability.
A properly configured cap inspection system covers five families of defects.
These are the most common defects and directly affect container tightness:
The tamper-evident ring shows consumers that the container has not been opened. The system detects broken bridges before consumption, partial breaks, or rings that are hanging loose. These defects can lead to customer complaints even when the product itself is in perfect condition.
European Union legislation requires caps on plastic bottles of up to 3 litres to remain attached to the container. This requires verification of:
These originate during cap manufacturing and include flash, microcracks, missing material, deformations, and, in metal caps or aerosols, dents. Since these defects are small and often have low contrast, they require high-resolution imaging and diffuse or low-angle lighting.
The system reads (OCR) and verifies (OCV) the batch number, expiry date, or DataMatrix codes printed on the top or side of the cap, checking that they are both legible and correct. An illegible code can compromise the traceability of an entire batch.
The cap inspection system is installed downstream of the threader or capper and operates on the moving conveyor. Four stages can be considered:
Traditional machine vision is based on programmed rules: contrast thresholds, edge measurements, or comparison against a reference pattern. It is fast, accurate, and highly reliable for stable geometric measurements, such as cap height or presence. Its limitations become apparent when the defect is variable or the environment changes: reflections on metal caps, different colors, condensation, or the hinge of a tethered cap can generate false rejects and require parameters to be continuously readjusted.
Deep Learning learns from labeled images of good and defective parts. It recognizes complex anomalies, such as microcracks, partially broken tamper-evident rings, or irregular flash, while tolerating the natural variability of the product. It also simplifies format and product changes.
The most robust architecture is a hybrid approach: traditional machine vision for measuring and verifying dimensions, combined with convolutional neural networks (CNNs) to classify complex cosmetic defects.
Integrating both inspections into a single station makes it possible to use the same image capture to check both the closure and the fill volume. Using backlighting, the system sees the liquid contour against the light and measures its height. This is the natural solution for transparent containers, such as PET bottles or clear glass. It detects underfilled and overfilled containers, while also being able to distinguish foam in carbonated beverages. For opaque containers, it can be complemented with X-ray or infrared sensors.
This combined inspection reduces line space and equipment costs and also makes it possible to correlate defects: a low fill level together with an improperly tightened cap often indicates a problem in the filler or capping head. All of this can be achieved without adding production stoppages or additional rejection points.
In PET bottles, glass containers, cartons, and jugs for water, soft drinks, oils, dairy products, and spirits, the system ensures tightness, preserves carbonation, and protects food safety at high production speeds. It also verifies batch and expiry-date coding required for traceability.
Dropper bottles, child-resistant (CRC) closures, and dispensing pumps require a perfect seal for patient safety and regulatory compliance. In luxury cosmetics, flawless aesthetic finishes are also required, with no scratches or marks. DataMatrix code reading is essential for serialization compliance.
In detergents, lubricants, and aerosols with integrated actuators or dispensers, a defective closure can cause hazardous substance leaks during transportation or use. Controlling safety closures helps prevent household accidents and costly claims.
Across all these industries, cap inspection using machine vision and AI protects consumers, reduces waste, and strengthens brand reputation with every unit produced.
At ATRIA, we have a team capable of developing customized automation and machine vision applications, from algorithm development to plant integration.
Some representative projects include:
This experience in surface defect detection, assembly verification, and container variability is directly applicable to cap inspection.
Would you like to learn more about cap inspection? Contact us!
Your success is our goal. We want to be your partner in the development of your projects, leveraging the power of innovation, flexibility, and the trust we provide. Together, we make a difference through new technologies.
Contact us and let’s build the future together!
