3 Key Technologies to Perform Quality Inspection with AI Machine Vision

In a fiercely competitive space, quality inspections are vital. A manufacturing and distribution enterprise may lead a market in its ability to promote its brand, close deals, and fulfill orders quickly and efficiently, but if product quality is lacking or inconsistent, customer satisfaction, loyalty–and the number of future contracts and revenues–will plummet.   

Although quality inspections have an essential role in a successful operation, finding effective ways to perform them can be a challenge. Manual inspections are time-consuming and often limit inspections to only a small representation of a run. Furthermore, inspectors may not perform checks consistently every time. They can become distracted by issues that arise throughout the day, or they may start a shift fresh but end up missing smaller defects as the day goes on.  

Some manufacturers have upgraded to an automated system with cameras that perform surface checks. However, these systems are not designed to perform comprehensive quality inspections, particularly for manufacturers that produce small, detailed products, such as passive components including resistors, inductors, and capacitors for electronic circuits. Manual methods or systems that perform only a cursory inspection often cannot detect subpar product quality that results from variations in temperature, humidity, or vibrations in those manufacturing processes.  

Also, quality inspections throughout the supply chain can be equally important to customer satisfaction and loyalty as inspections performed on the production line. Distribution centers must ensure products are packaged correctly, serialized if necessary, labeled correctly, and routed to the right pallets for transport, so customers receive accurate, on-time shipments.  

How to Make Quality Inspection Easier & More Advanced 

Machine vision gives machines, robots and autonomous devices the ability to see, detect and analyze images automatically, which can provide automated product quality inspection from the manufacturing floor, to warehouse fulfillment, to the distribution center. Modern machine vision systems leverage three technologies to perform better, faster and more cost-effective quality inspections. 

  1. Smart Cameras

Industrial, AI smart cameras, like our multi-award winning NEON Smart Camera, integrate AI capabilities directly within the camera itself by combining hardware with a pre-installed software environment.  

The all in one integration of image sensors modules, GPU or VPU modules, cables, industrial rich I/O, protocol communication and analytics improves compatibility, speeds up installation and minimizes maintenance issues, ideal for various quality inspection applications. 

Figure 1: NEON AI Smart Camera with pre-installed Edge Vision Analytics
  1. Artificial Intelligence (AI) 

AI adds a new dimension to quality inspections. AI-powered machine vision systems not only find defects or confirm proper packaging and labeling, but they can also make decisions based on context. The results can be dramatic; a study by McKinsey & Company found that businesses achieve 90% higher detect detection and 50% more productivity with AI.  

Standard machine vision is rules based, where a machine vision AI system becomes smarter as it consumes more images. Many factories and production lines are already using standard machine vision, which has the ability to detect when something is wrong but those systems cannot tell us what exactly is wrong (classification) nor direct a system to take an action once information is received. 

Figure 2: AI machine vision system deployed at a third party logistics company, Evans Distribution, ensuring accurate order fulfillment.

Automated quality inspection systems using machine vision AI software can classify what they see, become smarter over time and also create automation workflows. For instance, classifying a defect as a crease and sending a message to a conveyors controller to slow down. Machine Vision AI software gives machine vision experts the tools to build, test and deploy AI models faster while giving automation/IoT teams the tools to go connect, stream and automate (operationalize) machine vision work. 

Adding AI to machine vision hardware allows developers and system integrators to easily run different AI models directly in the smart camera to resolve many non-rules-based inspection challenges. Such as effectively finding defects or errors in different lighting, various positions, or when products are transparent or highly reflective to achieve a higher degree of accuracy, increasing productivity.  

  1. Edge Computing 

Edge computing is key to realizing the full value improved quality inspection with machine vision AI technology. As leaders in edge computing, our machine vision AI technology allows for real-time fast computing and AI inferencing right on the data producing thing such as a production line or piece of machinery.  

Edge intelligence enables systems to process a large volume of data without having to send it to the cloud, reducing latency and increasing efficiency. Here at ADLINK, our smart cameras are edge devices. So, take the NEON-1000-MDX, for example – it can leverage solutions such as the ADLINK Edge™ Software Platform or Edge Vision Analytics to process, analyze data and trigger immediate action so defects or errors can be addressed on-the-spot. 

Success Stories 

There are numerous examples of how smart cameras, AI, and edge computing together create machine vision AI systems that enable efficient, accurate quality inspections. For example:  

3PL company Evans Distribution Systems uses the ADLINK Edge Smart Pallet solution, comprised of two GigE cameras, an Edge AI Gateway and ADLINK Edge™ software. The system, which includes a model trained with VMLINK, records and classifies the contents of packages as they move through the line and shares data with Evans’ warehouse management system, providing automated quality assessments. The system is more than 99% accurate and frees employees to use their skills in other areas of the operation.  

John Deere is using an automated defect-detection solution for robotic arc welding. The system is an AI machine vision solution leveraging ADLINK’s EOS-i6000-M Series vision system featuring a 9th Gen Intel® Core™ i7-9700E processor, 4x Intel® Movidius™ Myriad™ X VPUs, and the Intel® Distribution of OpenVINO™ toolkit. The system, which sends a stop command in real time when it identifies a weld porosity defect, is 97% accurate.  

Figure 3: edge AI machine vision system for contact lens quality inspection

Passive component factories are benefitting from our Single Latch Activated Multipoint Comparison inspection system. It allows a single sensor to trigger multiple cameras, eliminating the cost and maintenance hours required for multiple-sensor systems. The system requires only one feed sensor to record the initial component position, and then when the component reaches the next inspection point, it automatically triggers the next camera. The system’s ROI includes real-time visibility, lower costs, and more efficient use of labor and resources.  

Have Confidence in the Products You Ship 

Product quality and accuracy on the manufacturing color and fulfillment center are table stakes for companies vying for market position. Manual processes and legacy systems that provide surface inspection capabilities offer only a limited ability to confirm that products and packages are free from defects or errors. Using machine vision systems that combine smart cameras, AI, and edge computing automate the process of performing a thorough quality inspection and confirm, without a doubt, that you’re meeting your customers’ expectations.  

Author: Chiawei Yang
Author: Chiawei Yang

Head of Edge AI Vision & IoT Products, ADLINK Technology


  1. Ich überlege Qualitätsprüfungen von Machine Vision durchführen zu lassen. Interessant, dass Machine Vision autonomen Geräten die Möglichkeit gibt die Bilder automatisch zu erkennen und zu analysieren. Da dies jeden Produktionsschritt begleitet, wäre das sicher gut für mein Produkt.

  2. Thank you for sharing this wonderful Blog. It’s a really helpful and Informational blog. Keep sharing technical & informational content. We are also from the same domain. We offer an Automated Quality Inspection System in Canada.

  3. Hi

    Quality testing & inspection is very good in this blogs , so many useful and unique information. Please share more software testing blogs .Great work and thanks for sharing the articles .

Leave a Reply

Your email address will not be published. Required fields are marked *