This website uses cookies primarily for visitor analytics. Certain pages will ask you to fill in contact details to receive additional information. On these pages you have the option of having the site log your details for future visits. Indicating you want the site to remember your details will place a cookie on your device. To view our full cookie policy, please click here. You can also view it at any time by going to our Contact Us page.

Close

Deep learning solves a difficult inspection task

13 August 2023

A health science production facility has employed a cloud-based Deep Learning system to solve a difficult application to inspect containers for the presence of transparent measuring spoons.

Nestle manufactures products for people with specific nutritional requirements, such as sip feeds and supplements, at its plant at Osthofen in Germany. Towards the end of the manufacturing process, a measuring scoop is inserted into each container prior to automated powder filling. 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.  

While the change to an almost colourless plastic scoop improved the recycling rate, the new transparent scoop, with its slight grey appearance, was difficult for the conventional image processing system to detect on top of an aluminium foil lid of a similar colour, which was also corrugated, embossed and reflective.

An intuitive solution
A solution was provided by SICK Deep Learning technology. Now operators follow an intuitive graphic interface to select and train their neural network in a few simple steps: Once the devices are set up, users are prompted to gather images of the inspection in realistic production conditions, and then sort them into classes. The pre-sorted images are uploaded to the Cloud, where the image training process is completed by the neural network.  The operator can then apply further production images to evaluate and adjust the system. When satisfied, the neural network can be downloaded to the Deep Learning-enabled device, and the automated inference process will begin with no further Cloud connection necessary. 

The 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 a SICK 2D camera. 

The image inference is carried out directly on the device in a short and predictable decision time, without the need for an additional PC, and results are output to the control as sensor values.  Whenever the 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.

Deep Learning can radically reduce set-up times and costs by enabling Artificial Intelligence (AI) image classification to run directly onboard smart devices. With Deep Learning, programmable devices take decisions automatically using specially-optimised neural networks and run accurate and reliable inspections that would have previously been challenging or simply impossible to achieve in high-speed automated processes.


Contact Details and Archive...

Print this page | E-mail this page