Drives and Controls Magazine
Home
Menu
‘First smart camera with deep learning’ tackles complex tasks
Published:  23 April, 2020

Cognex has announced what it describes as the world’s first industrial smart camera to be powered by deep learning. The self-contained In-Sight D900 embedded vision system incorporates the company’s ViDi deep learning software and is aimed at complex in-line inspection applications such as optical character recognition, assembly verification and defect detection.

The system is said to combine the self-learning ability of a human inspector with the robustness and consistency of a vision system, and to expand the limits of what can be inspected in factory automation. It can be set up using a few sample images. It uses Cognex’s easy-to-use spreadsheet platform and does not need a PC or deep learning expertise to deploy.

“Sophisticated manufacturers are increasingly turning to deep learning vision to solve inspections that are too complicated, time-consuming, or costly to program using human or rule-based machine vision,” explains Cognex’s senior vice-president of vision products, Joerg Kuechen. “By embedding our ViDi deep learning software on In-Sight, customers can solve even the most complex visual inspections quickly, easily, and more cost-effectively.”

Cognex's new smart camera uses built-in deep learning to solve complex in-line inspection challenges