11 February 2021
A sharp eye and a super brain: how AI is set to transform machine vision
Though interchangeably used by some, machine vision and artificial intelligence play a very different role. Combined, they are capable of achieving greatness for industrial operations – in particular for those where quality is made up by a complex system of rules and subtleties.
Machine vision is not a new technology; in fact, it has been around since the 1930s. Technology that preceded modern-day machine vision even dates back to early Egyptian optical lens systems (700 BC!) and the introduction of punched paper cards in 1801 (Joseph Marie Jacquard, for the automation of weaving intricate patterns). More recently, the application of machine vision found its purpose in straightforward manufacturing tasks where image capturing was linked to some form of software, and possibly robotics for the performance of a simple, repetitive task.
About machine vision
In isolation, machine vision is quite frankly useless. It will always rely on a software program for instructions. Machine vision is essentially all the hardware components that, when combined, provide an ‘eye’ within an operational process. The choice of these depends on the environment and desired outcome, but typically include one or more cameras, light sources and a processor. Machine vision is the ‘eye’, but needs the brain (the software) to interpret the image. This was limited to laborious and inflexible ‘hard-coded’ rules, until AI came along.
AI as ultimate brain power
From self-driving cars to voice recognition, artificial intelligence, or AI, has already found its way into society. Its application can be as broad as our imagination and seems only limited by the physical hardware it can tap into. In the case of industrial excellence, pairing AI with machine vision provides a highly advantageous proposition: a sharp eye on the process, feeding the images to an AI model trained to detect what the user wants to know. The two make such a strong team, that defects normally invisible to the human eye can be detected. Equally, the AI model can be trained to ignore inconsequential discrepancies (such as a fingerprints or dirt that do not affect the product). The possibilities are endless.
An additional benefit AI brings is its ability to continue to learn and improve. Further embedding defects to enhance quality output and accuracy is made possible thanks to the ‘human-like’ pattern recognition of these models. AI is the brain-genius that, when combined with machine vision, opens the door to new levels of operational excellence – and it is destined to transform manufacturing in the years to come.