A New Method to Inspect Flat tube Heat Dissipation Device for LED Lightings by Combining Machine Vision and Deep Learning

Abstract

Background/Objectives: With the advent of the 4th Industrial Revolution, factory automation technology including product inspection has emerged as a key task in the production of LED lighting.

Methods/Statistical analysis: It is very important to inspect the flat tube heat dissipation device, which is a key component, in the production of LED lamps in company L. Conventional rule-based algorithm is difficult to apply to inspection of flat tube heat dissipation device that require accuracy. In this regard, we propose a new inspection method using a production informatization algorithm that combines machine vision and deep learning.

Findings: The new inspection method using the production informatization algorithm has achieved a breakthrough in reducing the defective rate of 6% to 2% in the actual process. This is equivalent to an annual sales increase of 950 million won.

Improvements/Applications: The method presented in this study is a good benchmarking example for related companies because it can achieve the optimal effect without incurring high costs.

 Keywords: Machine Vision, Deep Learning, Production Informatization Algorithm, LED Lighting, Flat tube Heat Dissipation Device.