The project was based on SICK Inspector83x smart vision cameras utilizing Artificial Intelligence to automate quality inspection and product identification.<\/p>\n
The solution was built around three core AI functionalities.<\/p>\n
<\/p>\n
AI Classification<\/b><\/p>\n
The system automatically classified products and identified their variants, enabling:<\/p>\n
<\/i>Identification of different product types<\/p>\n
<\/i>Classification of products as OK or NOK<\/p>\n
<\/i>Verification of correct component selection<\/p>\n
<\/i>Detection of mixed product variants on the production line<\/p>\n
<\/p>\n
Object Detection<\/b><\/p>\n
The system localized objects in real time, enabling:<\/p>\n
<\/i>Product counting<\/p>\n
<\/i>Presence verification<\/p>\n
<\/i>Product position detection<\/p>\n
<\/i>Identification of missing components<\/p>\n
<\/p>\n
Intelligent Inspection<\/b><\/p>\n
The Intelligent Inspection function evaluated product quality by detecting:<\/p>\n
<\/i>Missing components<\/p>\n
<\/i>Incorrect product orientation<\/p>\n
<\/i>Quantity discrepancies<\/p>\n
<\/i>Colour deviations<\/p>\n
<\/i>Missing labels<\/p>\n
<\/i>Incorrect component positioning<\/p>\n
Before system commissioning, a reference image dataset was prepared and individual product features were annotated. Based on this dataset, AI models were trained to automatically recognize products during production.<\/p>\n<\/div>