23 Jul 2024


Camera with onboard deep learning tackles difficult apps

The intelligent inspection deep-learning app is designed to run on Sick’s Inspector P621 2D machine vision camera

Sick has launched its first machine vision camera with a pre-installed deep learning app, making it easy to create custom inspections of complex or irregular-shaped goods, packaging and assemblies, especially those that have previously defied automation.

The Sick Intelligent Inspection Deep Learning App runs on the company’s recently-launched Inspector P621 2D programmable camera. The combined package allows machine-builders and end-users to set up vision classifications using AI (artificial intelligence) in a fraction of the time and cost needed to program traditional vision systems to perform challenging inspection tasks based on recognising preset rules and patterns.

The new system can tackle applications where it was previously difficult to achieve consistent, repeatable quality inspections. It makes automation practical and affordable for complex tasks such as sorting fresh fruit and vegetables, checking the orientation of timber profiles by recognising the annual ring structure, checking leather car seats for creases or flaws, or inspecting the integrity of solder in surface-mount assemblies.

“By embedding the Intelligent Inspection App onto Sick’s Inspector P621 deep-learning camera, Sick has made it possible for users to purchase a ready-made package that uses artificial intelligence to run complex vision inspections with ease,” says Neil Sandhu, Sick’s UK product manager for imaging, measurement and ranging.

Users are guided through an intuitive process that teaches the system to recognise “good” and “bad” examples using cloud-based neural networks. Using a built-in image capture tool, they start by collecting example images of products in realistic production conditions. They are guided step-by-step via a graphic interface, and prompted to sort the images into classes. The pre-sorted images are uploaded to the cloud where the image-training process is completed using a neural network. More production images can be used to evaluate and adjust the system.

Once the user is satisfied, the custom-trained deep learning system is downloaded to the camera where it can start to take decisions automatically without needing further cloud connections. The results are output to the control system as sensor values and digital I/O.

Because the image inference is carried out on the device, there is no need for additional PCs. And because the system training is done in the cloud, there is also no need for separate training hardware or software, cutting implementation time and costs.

“Because it runs directly on the camera, the Sick Intelligent Inspection app does not require any additional hardware,” Sandhu explains. “So, users can automate complex vision inspections for a much lower cost of ownership. They can now consider automating quality inspections of products or goods that have just proved too difficult previously.

“Even better,” he adds, “the system can be set up in no time at all. Many users will be able to manage this process themselves. However, if needed, Sick is also offering services to support customers through the feasibility, commissioning and neural network training process.”


Users also have access to traditional machine vision tools installed on the camera, so they can extend the functions of their quality inspections. Developers working in Sick’s AppSpace can also use its Nova software tools for further custom development and to create their own SensorApps.

The Sick Inspector P621 is a 2D CMOS vision sensor with a 1.3 Mpixel image resolution. Illumination is provided by on-board LEDs and the IP65-protected camera has an adjustable electric focus. Its small size (71 x 43 x 35.6mm) means that the camera can be installed in tight machine spaces. It is compatible with EtherNet/IP, EtherCat, Profibus and Profinet.

The deep-learning technology is available as licensed option for all of Sick’s InspectorP600 2D vision sensors. Initially available with image classification, the app will be extended later in 2021 to incorporate anomaly detection, localisation and segmentation functions.

SickTwitter  LinkedIn   Facebook