Our customer manufactures printed circuit boards, which were manually inspected by two people in a three-shift operation due to the variety and complexity of the production. A classic inspection system was already used to check individual components – but a hundred percent automated end-of-line inspection did not happen, yet. Maddox AI has developed a semi-automatic inspection system in which printed circuit boards are inspected with a high-resolution camera. Due to the high product variability, a fully automated inspection system was not possible. The system is used to detect, among other things, defectively mounted components and other damages such as scratches, or solder bumps. By reading a data matrix code, the Maddox AI system tracks defects to a specific component. In the first step, the system is loaded manually. New defect classes have already been retrained several times free of charge.