The optical quality control of products and semi-finished products is a labor-intensive and therefore cost-intensive task. With the assistance system “imageSAM”, the Fraunhofer Institute for Process Engineering and Packaging IVV has developed a solution that allows companies not only to achieve 100 % production control but also to address the shortage of skilled labor in food industry.

By Dr. Isabell Rothkopf, Dr. Lukas Oehm

Chocolate production is a traditional craft. In order to economically produce large quantities for the food retail, support from production systems is necessary. These chocolate production systems have a long service life, often operating for decades. However, technical progress in the field of automation and digitalisation is rapid. This can mean that the systems quickly become outdated. Shift managers also report that experienced personnel would have to be deployed at various points in the plant to monitor production and intervene in the event of deviations. This is problematic because, on the one hand, it is uneconomical to have one person monitor all relevant production elements and, on the other hand, there is an increasing shortage of sufficiently trained and experienced personnel in the industry. In addition, not all areas are visible to personnel during ongoing production.

To solve this problem, the Fraunhofer IVV has developed the assistant system “imageSAM”, an automatic optical quality monitoring system that can be integrated into existing systems with minimal effort. Cameras are installed at the production steps to monitor and continuously record images of semi-finished products and products. The ”imageSAM” system analyses the captured images and classifies them using artificial intelligence (AI) according to the customer's specifications. This can be a differentiation between A-goods, B-goods or rejects. It is also possible to recognise different types of defects. The results are presented to the operating personnel on a tablet, smartphone, PC or a screen placed next to the system depending on the operator's preferences and integration options in the production workflow.

In the PralineDigital project, founded by IVLV e.V., “imageSAM” was cost-effectively retrofitted into an existing system and configured for the specific products manufactured there. As a result, the operator can view the camera image on his screen as well as the recognised defect classification, including suggested actions to rectify the cause of the defect. The AI-based fault detection achieves an accuracy between 91 and 98%, enabling faster identification and long-term solution of quality issues. Commissioning the system is no more time-consuming than training a new employee. Its simple integration into existing systems and user-friendly interface enable quality problems to be identified and rectified quickly.