Implementing Smart Factories in the Context of Visual Quality Control

Implementing Smart Factories in the Context of Visual Quality Control

Implementing Smart Factories in the Context of Visual Quality Control
07 September 2026

In today’s business world, both B2B and B2C customers prefer high-performance products rather than making choices based on a single criterion such as price, availability, or quality. Key factors driving this shift include the growing number of manufacturers, the development of new production technologies, continuous improvements made by manufacturers, and the resulting competitive environment. As a result, both product quality and process quality have evolved from being a competitive advantage to becoming a “necessity.”

Adopting the Industry 4.0 concept brings numerous benefits to smart factories through the digital tools it offers. Digital quality control improves the process by using cameras controlled by specialized hardware and software. Such control can be achieved through machine vision and classification algorithms.

As an example application, we took our first step toward machine vision-supported quality control on our rotary switch production line. In this initial trial, we implemented a cost-effective and easy-to-implement system, thereby achieving a reliable quality control process. A Raspberry Pi-based system was integrated into the movable terminal mounting process to verify the quantity, shape, and arrangement of the movable terminals. Images of the rotary switch housing are captured by a camera using OpenCV libraries. The captured images are processed to determine the RGB values at specified feature points on the body. A thresholding method is applied to detect the presence of metal contact points by comparing the contrast between the colors of metal and plastic parts. An integrated, rule-based classification algorithm classifies rotary switches as “acceptable” or “unacceptable.” An unacceptable switch can be sorted out based on the error code provided by the classification algorithm for subsequent processing.

This application has helped eliminate the need for one staff member in the process and improve process quality. Additionally, an abstract of the paper we prepared for the International Engineering Symposium (IES’20) was presented and included in the abstract book available at http://ies.idu.edu.tr/home-page/. This basic application can be further developed by taking into account requirements such as production speed, image resolution, or processing power needs. The key point here is that organizations must continue to make efforts to leverage the tools of today’s industrial revolution—specifically big data and machine learning—in the context of this application.

NEWS & BLOG
Other News