At the PETnology Conference 2025, Sascha Baumanns, Head of Sales at INTRAVIS, showed how artificial intelligence (AI) is changing quality control in the plastics packaging industry - not as a vision of the future, but as a lived reality. After all, AI technologies are already being used in over 1,000 INTRAVIS inspection systems worldwide.
Sascha Baumanns' presentation was divided into four central application areas that show how AI supports our customers in concrete terms:
Object classification is a central element of quality control: our systems reliably recognize whether a product meets the quality requirements or is defective. This is where AI comes into its own – it learns from thousands of images and makes decisions based on the most subtle differences.
One example of what this technology can do is OCRX, which has been used as standard in our systems for years. It recognizes characters and numbers on a wide variety of objects – such as preforms, bottles, labels or closures – regardless of font, contrast or lighting. And thanks to our extensive image database and specialized training procedures, there is no need for lengthy retraining on site during installation. This saves time and increases operational reliability.
These many years of experience – we have been using deep learning for character recognition since 2014 - shows that INTRAVIS is one of the pioneers in the use of AI in industrial vision inspection.
The challenges here are many and varied: for example, transparent objects, changing positions and contact between products. Our AI recognizes these objects with up to 20 images per second, calculates their rotation and ensures that the inspection always takes place at the correct position in the inspection system. This increases inspection accuracy and reduces false rejects – without any mechanical separation.
The strength of the AI is particularly evident with complex geometries, such as tethered caps, a wide variety of colors or unusual object shapes: it learns almost independently what is “normal” and reliably detects when something deviates from it. The evaluation takes place in up to 20 milliseconds and is therefore also suitable for high-speed lines.
With the IntraRecommender, INTRAVIS offers an intelligent production assistant that makes live recommendations for process optimization based on quality and process data. If an inspection system detects an increased error rate, the IntraRecommender suggests specific measures – directly to the machine operators.
The system continuously learns from operator feedback and automatically adapts its recommendations. In combination with cloud-based data storage, this
information can be accessed worldwide – a real advantage for internationally operating production sites.
Whether classification, object or anomaly detection or assistance systems – we at INTRAVIS use AI technologies in a targeted manner to make quality control more efficient, safer and more intelligent. For our customers, this means more process reliability, fewer rejects and a future-proof production.