The challenge
What needed solving
Operators classified 40 part variants by hand and entered counts into a digital device. Errors in classification and inventory tracking caused SKU mismatches and demand-supply gaps.
The solution
How Qualitas solved it
Controlled illumination and a uniform background help a camera capture each part. A deep learning model classifies type and color, counts parts, and sends results to a PLC.
Implementation
The team defined the 40 part variants, captured training images, validated classification and detection, and integrated the system with a PLC across inspection stations.
Results
The case study reports 98% classification accuracy, 93% object detection accuracy, and 91% color defect identification accuracy. Inventory data is recorded in real time in one second.
