FEATURE Machine vision
Visual inspection enters a new era with deep-learning systems
By Miron Shtiglitz, Director of Product Management, QualiSense S
imple rule-based algorithms have been used in the visual inspection market for many years, but as their limits have become more apparent,
the need for more sophisticated software has grown. Many of the most common solutions available in the visual inspection market can be characterised as rule based; i.e, they operate with diff erent algorithms and the parameters are set by an expert. For example, the rule might be to count how many black pixels are in an image, and if this is above a particular number fl ag this image as a defect.
Solutions like this are very eff ective at simple tasks, but for more complex problems, such as those in surface visual inspection, they are inadequate. Relatively small changes, like printer ink fading, varying lighting conditions, or even a minor change in the raw material would seriously impact the effi cacy of rule-based systems. Quality managers would need to call the service team out to continuously update the parameters. This requires sophisticated technology, hence the era of deep learning.
The first stage
Deep learning has now become a leading way of providing quality inspection. Such solutions can detect scratches or dents on surfaces which are poorly defi ned. By receiving many examples of defects, a deep- learning model can generalise the problem and identify defects on new parts by using this more general understanding. However, these models also require many training images, which must be entered manually – a lengthy and diffi cult process. The images must also be reviewed fi rst. For example, if the model fails to accurately detect a crack, more samples are required until the model understands what a crack is. Furthermore, simply providing the images is not enough, as some markup around the images is also needed.
It is important to bear in mind that the model can only be trained with images of defects. The end user must review thousands of images, identify the ones with defects, then carry out the correct markup before providing it to the machine. For most
24 September 2023 | Automation
deep-learning applications, hundreds if not thousands of examples are needed – just to get started.
Unsupervised learning: a new era? We are now standing at the beginning of a new, third era in quality inspection. Technological solutions aimed at overcoming the limitations of previous deep-learning models are generally referred to as unsupervised or semi-supervised systems. A key goal is to automate the process of model building described earlier. These models can be fed images of a sample of non-defective parts, the so-called “OK parts”. The initial model, whilst not perfect, will provide a basic understanding of what constitutes “OK”. It will use this generalised understanding to then fl ag suspected defects or outliers, and with feedback from the customer it will continually update and learn.
Automating the model-building process in this way is much easier for the user, but the biggest return on investment comes from the reduction in the time it takes to have an eff ective, working model. In this new era of unsupervised learning,
the process of model building can take place on the production line. Previously, the process of data categorisation would require one person working on their own for days before returning to the production line with a model. Often, models would not work as intended straight away, requiring
further training, which adds interruptions to the production.
In this new era of unsupervised or semi- supervised learning, the initial training mode can take place on the production line and can be complete in just 24 hours before it is ready to use. A semi-supervised solution is arguably superior to a fully-supervised one, since the outcome is a system that better understands the product. Until now, many of the people working on training the model were experts in data and AI but lacked a good understanding of the product. In this new era, semi-supervised systems will leverage the knowledge of the production people, and it will be their feedback that optimises the model. A fi nal advantage for those willing to embrace this new era is the capacity for diff erentiation between types of defects. Previously, the only goal was detecting a defect, with classifi cation being less important. However, more advanced deep-learning solutions can cluster together diff erent images and build an understanding into diff erent defect types. The data gathered from this can then help support both preventive and predictive maintenance.
https://lean-ai-tech.com/ CONTACT:
QualiSense
automationmagazine.co.uk
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50
