In our latest innovation, Cognex Offers a vision system that automates error detection in minutes – without PC or programming skills.
Increasingly, packaging products require their own customized inspection systems for optimum quality, eliminating false rejections, improving productivity, and eliminating return risk. Some applications of foundational machine vision along the packaging line include checking the presence, correctness, straightness, and readability of the label on the package. Other simple package checks include presence, position, quality and readability on the label.
But packages such as bottles, cans and boxes cannot always be checked accurately by conventional machine vision. For applications that present variable and unpredictable defects on confusing surfaces such as those with highly embossed or specular glare, manufacturers have typically relied on flexibility and judgment-based decision making by human inspectors. However, human inspectors have some very big trade-offs for the modern consumer packaged goods industry: they are not necessarily scalable.
Deep learning expands the range of possible screening applications
For applications that resist automation but require high quality and high throughput, deep learning technology is a flexible tool that application engineers can trust as their packaging needs grow and change.
Cognex’s deep learning technology can handle all different types of packaging surfaces, including paper, glass, plastic and ceramic, as well as their labels. Whether it’s a specific defect on a printed label or the cut-off area of a piece of packaging, deep learning solutions can identify all of these areas of interest simply by recognizing the changing appearance of the target area.
Using a set of tools, Deep Learning can then locate and compute complex objects or features, detect anomalies, and classify said objects or even entire scenes. Finally, it can recognize and verify alphanumeric characters using a pre-trained font library.
Simple solution, even for complex tasks
While manufacturers realize the importance of digitizing their processes with AI, many are still reluctant to invest in it due to a lack of resources. However, the combination of machine vision and deep learning is the way for companies to adopt smarter technologies that will give them scale, accuracy, efficiency, and financial growth for the next generation.
A new, full-featured vision system now puts the power of deep learning-based image analysis into an easy-to-use package that runs error-resistant applications in minutes.
The In-Sight 2800 system can be trained using just a few images to automate everything from simple pass/fail checks to advanced classification and sorting – no computer or programming required. The interface guides users through the application development process step-by-step, making it easy for even users with a new vision to set up any job.
Changes in products, materials or line speed? No problem!
The combination of deep learning and traditional vision tools gives users the flexibility to solve a wide range of screening applications. Tools can be used individually for simple functions or linked together for more complex logical sequences. A powerful classification tool can be trained using as few as five images to identify defects, sort them into different categories, and correctly identify parts with discrepancy.
The new In-Sight 2800 also offers a variety of industry-changeable accessories and components to help users adapt quickly to changes such as new parts, faster line speeds and higher quality standards.
Watch this video to see why the In-Sight 2800 is the easy choice for the next machine vision post, and enter for your chance to win the In-Sight 2800:
This content has been sponsored by Cognex.