Challenge
The client builds machines for pharmaceutical packaging. It wanted to detect defects more reliably while increasing production rates - in an environment where every inspection must be documented and compliant.
Solution
- Equipping the machines with machine vision for inline inspection.
- Combining classic image processing and machine learning - each where it works best.
- Designed for the line’s full speed.
- Compliance with GMP and the US Code of Federal Regulations.
- Commissioning, testing and documentation on the target system.
We use the same approach in other series production projects, with full analysis of every product during production. In port logistics, a segmentation model we trained replaces manual identification of cargo types at around 95% efficiency.
Result
- Higher defect detection at full line speed.
- Compliant, documented inspection instead of sampling.
- Reusable building blocks for further inspection tasks.
* Project: we owned the delivery
