Export-Oriented Integrated Knit Fabric and Apparel Manufacturer
A large integrated plant producing for leading European apparel brands inspected an average of 1,200,000 metres of fabric per month on greige and dyed fabric inspection tables. On manual inspection machines, operators missed yarn faults under 0.5 mm and slight shade differences, so the fabric went to cutting. When cut and sewn defective panels were spotted at final inspection, they were scrapped, wasting fabric, sewing and energy costs.
- 4 high-speed industrial line-scan cameras with special back/front diffuse LED lighting were integrated into the fabric inspection tables.
- The AI model was deployed to record holes, oil stains, yarn faults and width/length elasticity deformations with millimetre coordinates (X/Y axis and metre stamp).
- Connected to the inspection machine's PLC, the system stopped the machine automatically at critical defect density and sent the defect coordinate map to the CAD/CAM cutting software, so defective areas were automatically excluded from the marker plan during spreading.
| Metric | Before | After | Change |
|---|---|---|---|
| Fabric defect detection accuracy | 68.5% (manual) | 98.4% (AI-based) | +29.9 point increase |
| Defective fabric reaching cutting | 4.2% of total | 0.3% of total | 92.8% reduction |
| Inspection table speed | 18 metres/minute | 40 metres/minute | 122% speed increase |
| Customer return / penalty costs | ~€85,000 per year | €0 (6-month period) | Full success |











