How Machine Learning and Machine Vision are bridging the gap between reality and the future.
In a recent episode of our podcast, experts Maarten de Munck of Kapernikov and colleague Tom Roussel discussed the fascinating field of machine learning and machine vision. They highlighted what these concepts entail and their potential applications in different industries. In this blog, we provide a short summary of their conversation, heard in the podcast episode "Self-learning production systems that make their own decisions, future or reality? (Spotify).
Machine vision and machine learning in brief
Machine vision is interpreting images to make informed decisions based on them. This can be achieved using classical methods, such as label checking, without the need for machine learning.
On the other hand, machine learning uses sample data to automatically train a model to recognise patterns and make decisions. It has a wide range of applications, including image recognition and personalisation algorithms.
Machine learning: A form of artificial intelligence
Machine learning is a component of artificial intelligence (AI) that allows computers to learn and make decisions based on data. By training computer models, they adjust their decision-making processes and become "smarter" over time. Machine learning algorithms can determine, for example, whether an object is defective or whether a person is visible in a camera image.
In the podcast, the experts emphasised that these models focus on specific tasks and are not capable of general intelligence. Or as Tom points out, "machine learning systems are good at repetitive tasks, but they are not really smart. And that (intelligent -red.) piece requires a human operator who knows the production process inside out. That remains essential there."
Real-world applications
The experts shared several real-world examples where machine learning and machine vision systems have proved their worth. One such application involved detecting available space on shelves in a department store. By training a model to recognise boxes on the shelves, the system can alert an operator so that inventory management can be optimised. These technologies can be used in a variety of environments, from industrial settings to retail shops.
Advantages and impact
Machine vision and machine learning technologies have the potential to bring about major advances in industries and improve sustainability. They excel at repetitive and mundane tasks, improve efficiency and reduce human error. By automating certain processes, companies can create more attractive and fulfilling roles for employees, using their expertise in more valuable ways.
As Maarten points out, "Difficult jobs, tedious jobs, but also those that are sometimes very stressful because of ergonomics. For me, it would also be a piece of sustainability that we can use people in the best possible way. That we don't mess up our bodies during work, so to speak, that people can do their jobs without ailments, pain or injuries. I think those are also all things that our society needs." In addition, these technologies also contribute to waste reduction, quality control and environmental sustainability.
Accessibility and affordability
Implementing machine vision and machine learning systems is becoming increasingly accessible. Standard models and tools are readily available and recent developments have reduced hardware requirements. Although some expertise is needed for optimal results and high accuracy, even small investments can yield significant improvements. Companies can develop them in-house or use services from specialised suppliers.
Conclusion
Machine learning and machine vision offer an enormous potential to improve decision-making, optimise processes and enhance employee well-being. By harnessing their power, industries can achieve greater efficiency, sustainability and quality control.
This article is only a very brief summary of the whole conversation. Want to know more about the topic, the challenges, concrete examples and the future of machine learning? Then listen to the podcast (Dutch only) via:
