In the food industry, consistent quality determines shelf space, margin, and brand trust. A burnt, deformed, or underweight cookie in a delivery quickly triggers complaints and costs real money — through credit notes, replacement deliveries, and the disposal of entire batches.
Status quo before the change: A rule-based camera system was already in use. It worked with fixed thresholds and deterministic filters, but could not cope with the high variance of real defect patterns (color gradients, partial burns, irregular cracks, slight deformations). The result was high false alarm rates and frequent borderline cases. Parameter adjustments could only be made by the system manufacturer — involving lead times, service visits, and additional costs. In daily operations this led to expensive follow-up costs: unnecessary pseudo-scrap, manual rechecks during alarm series, short-term cycle time reductions, and occasional defective pieces that still made it into packaging despite the rules.
Image evaluation now runs locally at the line; the AI is trained with representative good and defective samples and detects color deviations (too light/too dark), burns, cracks, broken edges, or deformations within milliseconds — robust against natural variance in shape, color, and crust.
Inaccuracy: Fixed thresholds and filters led to high false alarm rates, borderline cases, and defective parts slipping through despite camera use.
Precision: Every product is reliably inspected, with color deviations, burns, cracks, and deformations accurately detected.
High costs: Parameter adjustments only possible via the manufacturer — causing service visits, extra costs, and delays.
Cost reduction: Decisions are track-specific and reproducible, service dependencies are eliminated.
Production issues: Pseudo-scrap, manual rechecks, and cycle time reductions burdened quality and efficiency.
Production stability: Fully automated rejection without line stoppage, fewer complaints, and higher efficiency in shift operation.
Find out in detail how our customer increased its product quality and saved inspection costs at the same time by using an AI-based visual quality inspection.
© 2025 ANTICIPATE GmbH. All rights reserved.
