Deep learning, one of the most exciting developments in the field of machine learning, has the potential to change the game in industrial image processing.
Inspecting rain sensor components with complex, reflective, and streaked surfaces had been a long-standing challenge for our customer. Implementation using conventional image processing proved difficult, which ultimately led to the introduction of manual inspection.
A network architecture specifically tailored to this task enabled us to accurately distinguish between flawless and defective rain sensors — even with just a few hundred training images.
Die daraus entstandene Prüfautomation integriert sich perfekt in den Produktionsablauf und kontrolliert täglich vollautomatisch 12 000 Regensensoren. Die kostenintensive manuelle Kontrolle konnte somit komplett durch das AutoLab Deep-Learning-Tool ersetzt werden.
More information about this project can also be found in our LinkedIn blog post.
