Neuronal competence for over 25 years

In the sense of Industry 4.0

Deep Learning is about to turn into a key technology for the manufacturing industry. Already, automated quality control using image processing in particular shows remarkable progress. For 25 years, NeuroCheck has been using self-developed back propagation networks for the classification of image content and object characteristics. The necessary neural network structures have been an integral part of the NeuroCheck image processing software since 1994. With increasing computing power, cloud computing and fast Internet connections, entirely new possibilities for the use of deep learning in industrial vision open up today. Inspections tasks that are nearly impossible to solve thus become feasible. The current version of NeuroCheck includes the latest network models for your inspection tasks.

NeuroCheck as a Platform » The Basis for your Deep Learning Application

You already have deep learning know-how and have developed your own models? Now you would like to integrate the results quickly and easily into your production process?

NeuroCheck provides you with all the necessary interfaces for your endeavor. Neural networks can be integrated directly into NeuroCheck in the TensorFlow format. Other model types such as CNTK, Watson, and others are also possible. As a basis for your deep learning application, process integration (Profibus, Profinet, EthernetIP etc.), hardware support (2-D, 3-D cameras), and data acquisition (image data, XML, data base) have already been part of the standard scope of NeuroCheck for a long time.

Main Areas of Application » Successful Deployments when it really makes sense

As a pioneer of industrial vision, we have been developing and using neural networks that have already been successfully integrated in modern deep learning applications. We only make use of this complex and computation-intensive technology when it is sensible and offers the prospect of success.

Classification

In many cases, objects or areas need to be classified, e.g. in presence verification or the identification of OK and NOK parts. In this example, the screw consists of a thread and a screw head. Using pre-defined data, a classifier learns the specific characteristics of the objects and classifies them accordingly.

Object Recognition

If there are several areas or objects to be classified within one image, object detection can be used to localize and classify simultaneously. In this example: Screws of varying colors with varying threads and screw heads.

Example Application » Inspecting a Rubber Sleeve

False Alarm Rate reduced significantly

As part of an automated production step, a rubber sleeve is affixed to a carrier. This can cause pinches and back folding, which in turn may compromise leak tightness and thus the proper working of the assembly group. The challenging part of this inspection task is the high variety of fault appearances that can occur and that it is impossible to establish a quantitative classification of the faults. In addition, the component images differ in size, position, contrast, and brightness due to the production process. Traditional image procession produces a high rate of false alarms with an additional high risk for errors going undetected.

By using deep learning, the false alarm rate was reduced more than tenfold, thus the recognition rate was significantly improved.

Benefits of deep learning

• Robust with regard to environmental factors such as illumination, varying contrast and background
• Varying sizes and positions
• Broad range of fault characteristics

Example Application » Identification and Localization of Screws

Neural Network solves Inspection Task

To control an automated screwing process, precise identification and localization of the screws is necessary. The station processes a high number of various assembly groups with varying numbers of screws, surfaces and geometries. In addition, the screws may be slightly covered or absent, which must lead to a termination of the screwing process. With this application as well, a number of complex environmental conditions need to be dealt with. The example images show various levels of brightness, various sizes of assembly parts, positions and levels of focus.

Traditional industrial vision processes require the creation and adjustment of special inspection programs for each assembly type. The high number of variants made this approach very time-consuming and entailed maintenance difficulties.

By using a neural network, the entire task was solved with a single program. In addition, robustness of recognition under changing environmental conditions was greatly improved compared with the traditional process.

As your competent partner, we look forward to be of assistance

Our work in this area is based on deep learning and machine learning algorithms and ranges from basic research to industrial knowledge transfer to the implementation of deep learning systems.

We look forward to hearing from you

If you have any questions or require additional information regarding deep learning in industrial vision, do not hesitate to contact us using the contact form, via e-mail or phone +49 7146 8956-0. Our experienced experts will give competent and comprehensive advice.