Defect detection of plastic caps

Implementing AI Computer vision solution for plastic cap sorting.

BUSINESS PROBLEM

Our customer - plastic part producer, producing approximately 180 plastic caps per minute, reaching 3-5% defect rate in the production sold to customers. Manual defect detection proved to be inefficient due todefect sizes, color of the product and human errors.

PROJECT OBJECTIVES

  • Accurately detect 3 types of defects
  • Implement a solution, capable of detecting 180 parts/minute
  • Detect defects as small as 3mm size.

SOLUTION

Our advanced AI Computer vision solution solution was integrated into the existing production line without disturbing the existing setup. AI models were trained with hundreds of images for each defect type, to reach high detection accuracy.

AI models detect defects on a moving conveyor belt and send alerts whenever defects are detected.

Our engineers worked on a few challenges, including:

  • Detecting defects when caps that are overlapping
  • Detecting caps with extra material
  • Detecting air bubbles in the products

RESULTS

The entire solution was set up and deployed in 2 weeks.

Our solution reached overall accuracy rate of 99%, capable of detecting all 3 types of defects, resulting in 0.2% defect rate in the final production results.

Plastic cap defect detection process