FEATURE Machine vision


Feature sponsored by


Digging deep for automated inspections


Nathaniel Hofmann, SICK’s Market Product Manager for Machine Vision, outlines a simple way to fast-track to more challenging machine-vision inspections using Deep Learning


F


or products or scenes where there are potentially an infi nite number of variations, the expert advice has always been that machine vision


has its limits; some products, processes or components have been just too varied to teach to a machine. These barriers are being swept away by Deep Learning. Deep-learning machine vision mimics the way humans make sense of unpredictable visual data that can’t be classifi ed with a set of rules, thus giving it the skill to make judgements based on its knowledge of a highly-variable dataset. This will solve applications such as spotting defects or foreign objects, or classifying natural products based on visual appearance.


Examples versus rules Traditional machine-vision inspection tools use “if-then” rules, to compare images against a set of pre-determined parameters, whereas a Deep Learning system learns by being shown many real-life variations. This is known as an “example-based”, as opposed to a “rule- based” process. There is no need to select from the toolbox of algorithms used to identify defects, such as pattern fi nding or edge detection. Instead, Deep Learning cameras can automatically detect, verify, classify and locate objects or features by referring to the library of images that it has learnt.


Deep Learning can sort and classify


products or components that could be organic, like wood or food; easily deformable, like a plastic bottle; widely variable, like solder spots on an electronic component or glue spots on an automotive part; or they could be highly-refl ective, such as a metal part or a packaging fi lm.


Intelligent inspection With SICK’s Intelligent Inspection


28 September 2023 | Automation


SensorApp, which runs, for example, on the InspectorP 2D vision sensors, users begin by collecting example images of their product, package or component in realistic production conditions. Following step-by-step prompts, they teach the system to recognise examples as pass or fail.


The images are uploaded to SICK’s cloud-based training service, dStudio, where the training process is completed by specially-optimised SICK neural networks. The custom-trained Deep Learning solution is then downloaded to the vision sensor and the automated inference process can begin with no further cloud connection necessary. The SensorApp also off ers the


fl exibility to retrain machines when new products are added, adapt when processes are changed, and respond to a higher variety of items being produced at the same time.


Anomaly detection


SICK’s Deep Learning can be trained to recognise when there is an anomaly outside of the norm, while tolerating numerous acceptable variations in the objects being inspected. For example, it could be trained to recognise when a fi lm label is applied out of alignment on a bottle. Or it could be trained to detect when there is dust or a scratch on a refl ective dashboard display. Usefully, with anomaly detection, operators can then build up a picture in real time of a “heat-map” of defects, which can be used to identify and correct abnormal trends in a timely manner.


A healthy solution for Nestlé A Nestlé health science production facility in Germany is operating SICK’s cloud-based Deep Learning system to solve a diffi cult application to inspect


Nestlé detects plastic


spoon with a SICK Deep Learning system


containers for the presence of transparent measuring spoons. Nestle manufactures products for people with specifi c nutritional requirements, such as sip feeds and supplements, at its plant at Osthofen in the Rhinehessen region. Towards the end of a supplement manufacturing process, a measuring scoop is inserted into each container prior to automated powder fi lling. Until recently, as part of the quality control system, a vision camera using a colour pixel counting tool inspected for the presence of the plastic scoop in the container at a process speed of over 80 cans per minute. Changing to an almost colourless plastic scoop improved the recycling rate. However, the transparent scoop with its slight grey appearance was diffi cult for a conventional image-processing system to detect on top of an aluminium foil lid of a similar colour (see image above), which was also corrugated, embossed and refl ective.


At Nestlé, SICK neural networks


were trained by being shown images of the enclosed scoop in a wide variety of orientations. Then, the taught-in decision- making algorithm was downloaded to the SICK picoCam 2D camera.


Whenever the SICK Deep Learning system detects that a scoop is missing, it stops the system. As soon as it detects that a scoop has been added, it allows the process to continue without a manual restart.


Nestlé achieved extremely high


reliability from the system, while implementation was simplifi ed. The system also off ers Nestlé the fl exibility to expand it to run new applications.


CONTACT:


SICK (UK) www.sick.co.uk


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