Heuft has released updates to its InLine II IX system, combining pulsed X-ray technology with AI-supported image processing to improve detection in empty bottle inspection while reducing false rejection rates.
The system integrates in-house developments including Multi Color Image Processing (MCIP), HEUFT reflexx A.I. cameras and deep learning models. These were presented at drinktec 2025 and incorporated into the measuring device shown at the event.
MCIP, developed by the company, enhances sidewall inspection and enables its AI to detect defects that were previously not visible, including small cracks and faults in applied colour labels. The same technology now enables full inspection of the mouths of wide-neck bottles, using structured illumination and automated evaluation to identify small defects.
A deep learning model trained on images of the necks of 0.5-litre Euro beer bottles identifies deviations in shape, glass thickness, brightness and colour. According to the company, images outside defined tolerances are automatically classified as faults, allowing containers with critical or new defects to be removed while others are reused.
AI is also used to centre bottle bases for consistent inspection regardless of brand-specific characteristics. Additional pre-trained deep learning models assess the entire base image of both 0.5-litre Euro glass and 0.7-litre GdB PET returnable bottles, using anomaly detection to identify new errors.
The company says the combination of MCIP and deep learning delivers more stable and clearer results under varying environmental conditions, improving detection reliability.
Pulsed X-ray technology, combined with HEUFT reflexx A.I. X-ray image processing, is used to detect glass splinters and fragments on bottle bases and distinguish them from harmless structures.
