Case Studies

Metal Laminate Coils QC

Technologies

Computer Vision
Nvidia Jetson

Client’s Need

Automate end-of-line quality control of rolled metal coils, currently performed manually by an operator in steel production processes. Initially leverage CCTV data to customize and test AI computer vision algorithms capable of detecting product defects. Expand the number of defect categories to be monitored and their accuracy, possibly using other camera systems.

Approach

A solution capable of automatically detecting various defect categories, leveraging AI applied to computer vision. The trained AI algorithms recognize any asset defects (predefined categories of coil damage) based on the variation from a set of configurations (state) of damage-free assets (baseline) by change detection technology.

Main Outcomes

  • The solution is applicable for both in line quality control and end of line quality control.
  • Implementation suitable for both on the edge (i.e. Nvidia Jetson) and on cloud.
  • TRL 6, tested onboard machine with standalone Nvidia Jeston hardware target.