Do you want to incorporate deep learning models into your machine vision systems?
If so, you are not alone.
Deep learning is based on multi-layered neural networks that can identify patterns by analyzing data. These algorithms learn continuously as they review new data.
Integrators, machine builders, and end users are building deep learning algorithms into machine vision systems to help automate such processes as inspection, sorting, and assembly. The technology allows these professionals to solve problems that are resistent to classic machine vision approaches.
Related: Five Steps for Building and Deploying a Deep Learning Neural Network
In many cases, engineers combine both hard-coded programming and deep learning technologies for optimum results. For example, they may use a deep learning model to segment images but then use a classic machine vision algorithm to measure the dimensions of a part.
Automating Image Analysis
In fact, these newly automated processes often rely on the ability of deep learning to help execute the underlying image processing tasks. These include anomaly detection (which recognizes deviations), classification (which assigns objects to categories), object detection (which finds objects in an image), and segmentation (which divides images into regions of interest).
As Ulf Schulmeyer, a product manager at MVTec (Munich, Germany), explains in an article for Vision Sytems Design, the advent of deep learning has expanded the scope of processes that can be automated—ranging from grading fruit to inspecting semiconductors.
Related: How to Deploy Deep Learning Neural Networks in Machine Vision
For example, determining whether an apple passes inspection is challenging because there are so many naturally occuring variations size, shape, and surface color and texture. It would be nearly impossible to write code that covers all of the possiblities. However, deep learning models learn to predict what variations are acceptable even if they haven't confronted the exact situation before.
"The recognition rates that deep learning delivers allow machine vision systems to reach new levels of quality control. This also allows entirely new applications to be automated based on machine vision. Deep learning is a development that gives new impetus to the entire industry," Schulmeyer writes.
Related: What is Image Segmentation with Deep Learning?
As deep learning continues to mature, Vision Systems Design will provide you with the latest insights on trends, new algorithm types, applications examples and more.
Stay tuned at www.vision-systems.com or follow us on social media including LinkedIn, Facebook, YouTube, and X.
About the Author
Linda Wilson
Editor in Chief
Linda Wilson joined the team at Vision Systems Design in 2022. She has more than 25 years of experience in B2B publishing and has written for numerous publications, including Modern Healthcare, InformationWeek, Computerworld, Health Data Management, and many others. Before joining VSD, she was the senior editor at Medical Laboratory Observer, a sister publication to VSD.