In a joint assembly/quality Podcasts, Jennifer Piercinga multimedia editor with assembly magazines and assembly auditable hosts, and Michelle BangartManaging Editor for qualityExplore the area where assembly and quality intersect: machine vision, David Decaufounder and owner of Machine Vision Source. He will speak at the assembly show this fall. His session is “benefiting from machine learning in machine vision applications.”
quality/assembly: So you'll be speaking at a show in Rosemont, Illinois this October. Could you tell me a little about how you decided to talk about machine learning?
David: Well, first of all, I think there is some confusion about what machine learning is, what AI and deep learning is. They are all generally part of the same bucket, and machine learning is part of computers and data science, but it is very closely related to AI. Many people think about deep learning. This is something that most people find more familiar with, and considers deep learning to be part of machine learning. The problem is a very valuable set of algorithms within machine vision that helps you perform simple classification, simple segmentation, and other tasks without moving into a full-scale convolutional neural network that is part of a deep learning algorithm. That's good.
I know in the past that you're good at following trends, but as an engineer, you're too caught up in them and don't want to use something or use something just because people are talking about it. So. absolutely. And when we talk, when we think of vision engineers, when we think of machine learning, it's something that doesn't appear very often. So it's not one of those trendy topics, it's really, really worth it in a particular use case. And I look forward to being able to present these options to the audience.
quality/assembly: We mentioned machine vision. Is it the machine learning sector? I'm just trying to learn the right terminology.
David: That's right, that's right. And I shouldn't assume that anyone has such a broad term completely under their belt. We usually come into the definition of these things and many people think that machine vision is involved, and we think of derivatives from the early days known as computer vision. Today, many people pick up the term as a kind of catch-all of what is related to using computer vision to perform analysis.
However, machine vision appeared in the early days of AI, literally back in the 50s, and machine vision was not actually part of computer vision, but in the late 60s. Therefore, machine vision can be considered in a broader sense, as it considers computer vision. The other distinction now is that machine vision is mostly recognized to extract data from images using separate algorithms such as machine learning and many other related algorithms. Meanwhile, computer vision in today's market and in today's industrial environment is thought to define the use of deep learning to extract their features and segment their images.
