Normally, the quality monitoring of foodstuffs is based on technical parameters such as shape, moisture, color, weight, dimensions, degree of ripeness or level of browning. Safe-Ident Quality, on the other hand, checks non-quantifiable aesthetic aspects of a food product: The visual beauty.
Deep-learning processes enable the system to make judgements just as an experienced QM employee would. During production, it can thus already be ensured that a product pleases the critical eye of the consumer!
Features SAFE-IDENT Quality
• Sample inspection of raw products
• Fully automated reporting & archiving
• Deep-learning classification
• Customizable evaluation system • Avoids rejects
• Increases product quality
• Fully automated reporting & archiving
• Deep-learning classification
• Customizable evaluation system • Avoids rejects
• Increases product quality
Teaching the system
At first, experienced operators manually evaluate images of the product and select freely from the user-defined evaluation system, e.g.: school grades, quality classes, good-moderate-bad, IO-NIO. There is no need to justify or explain the judgement because no individual deterministic facts are evaluated in order to indirectly derive the quality, but the subjective overall impression is directly recorded. An output of the classification reasons in the form of a table is also possible. Based on deep learning methods, the system learns to assess the quality of a product in the same way as the quality managers.
Quality Inspection
During production, a sample inspection is carried out at regular intervals. The employees carrying out the inspection do not need to have any knowledge of quality assessment. The sample is inserted into the station, the recording is started with a click and the assessment is displayed directly by the system. A report is generated and archived.
One version of SAFE-IDENT Quality can be delivered as a complete station with an integrated camera. Our solution is alternatively available as an OEM version that can also be used inline.
