How does Deep Learning make an impact in Machine Vision?

David Dechow:
The thing about AI and we’ll more specifically target that as DL, deep learning, is that it handles imaging and handles the images a little bit differently than traditional machine vision, and allows us to do subjective analysis in a much better and an easier way than we have been able to in past years. When we have very subjective features that can only be identified by experts, and even then the identification is more of a human task, deep learning has really excelled at those types of tasks. I’ve done several in recent years, one that comes to mind is the ability of the deep learning technology to see variations in food products, variations in an individual food product that just can’t be quantified in basic rule-based imaging. Another one is where we have seen the ability of deep learning to quantify thermal images where the profile of the thermal image is very unique and can be justified by a human inspector, but not necessarily by rule-based imaging. And I think we’ll see that many of the use cases that struggle with discrete rules tend to be achievable by a deep learning solution.