Automated Physical Marking of Defects with Computer Vision

The challenge

The client already detected defective ceramic parts with computer vision but needed those parts marked automatically on the production line.

The approach

Kumulus prepared the data, developed and validated the models, and integrated the existing computer vision system with an embedded solution that acts directly on the production line.

The outcome

Defective parts on the ceramic production line are now marked automatically, supporting quality traceability and reducing manual intervention. The existing computer-vision inspection system was connected to an embedded water-spray marker, with detection and piece counts synchronised on a Jetson Nano.