Deep learning in depth

March 12, 2020
The answer to the question of “What is deep learning?” may depend on the person asking.

The answer to the question of “What is deep learning?” may depend on the person asking. 

If a person with a non-technical background asks, you might reply by saying that deep learning is a machine learning technique that trains computers to perform human-like tasks, such as recognizing objects in images. If a person with a more technical background who is less familiar with deep learning asks, an answer given by Perry West, Founder and President of Automated Vision Systems, Inc. ( on a recent Vision Systems Design webcast ( may be helpful: 

“Deep learning can be thought of as programming with data,” he says. “In a virtual machine setup with a neural network, the network is initially not programmed to do anything. When this virtual machine is given images, it begins to program itself to perform whatever the task is that is being set up.” 

Within the machine vision space, deep learning is a technology that has grown exponentially in the past few years. A recent survey of 320 Vision Systems Design audience members ( indicates that 56% of respondents currently use deep learning technology in some capacity. This breaks down to 31% using often and 25% using seldom. Still, not a whole lot of companies offer the technology as part of their current portfolio and knowing where to start can be daunting. To help provide a comprehensive guide on deploying deep learning in machine vision, this issue’s product focus (page 20) article looks at the various types of commercially available deep learning tools today. 

This includes open-source frameworks, specialized libraries, complete deep learning products, and deep learning integration services. In addition to defining each type of deep learning offering, the article provides several examples of products and the various capabilities offered by each of them. Two other articles in the issue also tackle deep learning, including an article on page 9 describing a new dedicated chip that enables efficient embedded device processing and another also on page 9 detailing a neural network for optical analysis that accurately retrieves phase information from 3D point-spread functions. 

Additional topics covered in the issue that we hope you’ll find informative and helpful include hyperspectral imaging, industrial inspection, real-time image dehazing, LiDAR, and 3D imaging, among others. If there are other topics, technologies, or applications that you believe we should be covering more often, we’d really like to hear from you. Whether it’s an article idea, an application you’d like to talk about, a new product you’re excited about, or any other questions or suggestions related to our coverage of machine vision and imaging, please reach out by sending me an e-mail at [email protected].

About the Author

James Carroll

Former VSD Editor James Carroll joined the team 2013.  Carroll covered machine vision and imaging from numerous angles, including application stories, industry news, market updates, and new products. In addition to writing and editing articles, Carroll managed the Innovators Awards program and webcasts.

Voice Your Opinion

To join the conversation, and become an exclusive member of Vision Systems Design, create an account today!