On a single production line, bottles of every shape, size, and label roll past at high speed. Unilever produces 89 different shampoo SKUs this way—one line, dozens of formulas, endless opportunities for something to go wrong.
Existing vision systems from vendors like Cognex and Keyence could only inspect one SKU at a time. That meant either dozens of separate programs or constant reconfiguration—slowing the line, increasing the risk of errors, and leaving operators frustrated. Scratches, dents, underfills, and misprinted labels were still slipping through.
A Smarter Approach
Elementary brought a different solution. Instead of creating one program per SKU, a single AI-powered model could watch the entire line. It detected defects across multiple SKUs, no matter the bottle’s orientation or label layout. The system also learned continuously, improving detection over time—letting operators focus on running the line, not fighting the inspection system.
What Changed
The results were immediate: fewer defects, smoother production, and operators who could trust the line. Key improvements included:
- Real-time defect detection: dents, scratches, misprinted labels, and underfills caught as they happened
- Reduced manual checks: fewer line stoppages and less repetitive work for operators
- Streamlined high-mix production: one system covering dozens of SKUs instead of dozens of programs
See the Full Story
Unilever’s experience shows what’s possible when inspection keeps up with the complexity of modern production—without slowing the line or adding unnecessary programs.
