The Challenge
A personal care products manufacturer in Bengaluru running two packaging lines at 480 units per minute was experiencing label and seal defect escapes to trade customers. Spot-check sampling caught only 30% of defective batches. Three retail customer complaints in a single quarter — including one partial recall costing ₹18 lakhs — triggered an urgent search for 100% inline inspection.
Deployment
Two inline inspection stations installed after the labelling machine and before carton packing on each line. DeepVision trained in 3 days on 1,800 images covering label presence, alignment (±2mm), batch code legibility, and seal integrity. Go-live was day 7. Validation accuracy: 99.1% on labels, 98.4% on seals. False positive rate: 0.4%.
Results After 4 Months
| Metric | Before | After | Change |
|---|---|---|---|
| Label defects per million (DPM) | 9,400 | 1,150 | −88% |
| Customer complaints (label/seal) | 3/quarter | 0 | −100% |
| QA inspection time per shift | 4.5 hrs | 0.5 hrs | −89% |
| Monthly relabelling cost | ₹4.8L | ₹0.4L | −92% |
“We went from three customer escalations in a quarter to zero in four months. The system also found a root cause in the labelling machine timing we would never have discovered from manual sampling.” — Plant Manager, FMCG Personal Care, Bengaluru