The client required to equip their inspection machines with an automatic system that could read the batch number code printed on the metallic curved surface of vials necks in dotted letters.
Quality control was being carried out through traditional machine vision approaches that were inadequate at managing the bad printing quality and distortions on the image due to the curved printing surface. The consequence was a high rate of false scrap since the system discard all the vial in which the lot number was not well printed (approximately 10% of the total production was incorrectly rejected)
The vision system already installed on the production line was re-used to acquire printed code images. These images are loaded directly into AI-go Studio and used for the models specialization. These models identify the individual characters, return the complete code reading and a feedback according to the data coming from the customer tracking system.
The trained and tested models were then deployed in production through AI-go Runtime, ensuring full compliance with the cycle times of the production line.
Increased OEE (Overall Equipment Effectiveness). Maintaining the same quality assurance, but reducing false rejects: the amount of false reject decreases to 0.5% on the total production.
Reduction of the time required for format change operations. Thanks to the user-friendly interface, less experience and time is needed to achieve good performance, even on the new formats.
Most reliable output. Better recognition of characters – even if badly printed and in suboptimal visual conditions.
Reduction of the time required to add extra quality controls already present on the line (e.g. OCV).