Customer challenge:
For window manufacturers, ensuring product quality, such as checking the foam filling of aluminum profiles, involves manual inspections. While this system works, there are clear limitations. Accurate component assessments require experienced, alert employees. And because staff has to work quickly and conduct the same assessment all day long, there is a risk of fatigue and becoming “speed blind,” which compromises the quality of the assessment.
The solution is a combination of Intelligent Inspection with SICK AppSpace SensorApps:
A 2D machine vision solution powered by deep learning checks that aluminum profiles of a window blind are accurately and sufficiently filled with polyethylene foam, and is able to distinguish holes in the filling from dirt. With dStudio and some example teaching images, it is easy to prepare a neural network for the SICK sensor. The software learns to distinguish between a homogeneous foam filling and an incomplete filling at the cut surface.

Benefits:
The solution is user-friendly and quick to learn. It offers combined benefits from traditional rule-based tools in tandem with deep learning that solves complex machine vision applications. Accurate automated inspection and production reduces waste and warranty claims and improves reliability and productivity. On-device application buildup is fast and easy, and accuracy and execution speed training can be optimized in the SICK dStudio.










