An industrial components manufacturer

Catching defects at line speed

Edge-deployed computer vision that inspects every part on three production lines, reducing escaped defects and inspection fatigue.

Defects reaching customers
−45%
Of parts inspected, up from sampling
100%
Payback period
11 mo

The challenge

Quality inspection relied on operators examining a sample of parts under magnification. Accuracy varied by shift, and small surface defects were sometimes missed.

Customer returns were rising, and the manufacturer wanted to inspect every part without slowing the line.

Our approach

  1. 01

    Data collection

    Industrial cameras captured over 60,000 images, which quality engineers labelled using an efficient annotation workflow.

  2. 02

    Model development

    A detection model was trained and optimised to run in under 30 ms per part on edge hardware.

  3. 03

    Line integration

    Results feed directly into the line PLC to divert suspect parts automatically.

  4. 04

    Continuous improvement

    Operator overrides are captured and used in monthly retraining cycles.

Results

  • Defects reaching customers fell by 45% in the first two quarters.
  • Every part is now inspected, at full line speed.
  • Operators were redeployed to root-cause analysis and process improvement.
“They were as comfortable on the factory floor as they were in a notebook. The system simply works, shift after shift.”
Plant Director · Industrial components manufacturer

Let’s build something that lasts.

Tell us about the problem you want to solve. A senior consultant will reply within one business day.