Home   Contact   Search de | en

Why Deep Learning

Detected error classes (from left to right): rolledin scale, patches, crazing, pitted surface, inclusion, scratches | © NEU Surface Defect Database

For complex tasks, traditional programming is much more complicated than deep learning

because of the algorithmic description of all possible variances. This makes classification tasks much easier to solve than with existing algorithmic methods.

Neural networks are particularly well suited for many other tasks, such as reflective surfaces, poorly illuminated environments, varying illumination, moving objects, robotics and 3D applications, where conventional methods clearly reach their limits and can only be realized with great effort and expertise.

Classical algorithms are still suitable when the localization of objects or
errors in an image, measurement checking, code reading or post-processing
are needed. Deep learning, on the other hand, offers very high reliability in
recognition rates and will improve the quality of today‘s image processing
systems and open up new applications.

Areas of Application

Deep learning is already in use today in sample and object detection with classification. The procedure achieves the best results with varying objects and identification of defects or anomalies as well as with difficult surfaces including transparencies and reflections. In the manufacturing process, a machine is thus able to manage a variety of variants even in varying surrounding conditions. Here, deep learning is used successfully in condition monitoring and predictive maintenance. Further areas of application encompass inline inspection, robotics, pick and place, autonomous driving and driver assistance systems, drones, satellite imaging, agriculture, medical technology, cellular research, and cognitive systems that work with humans, such as those used in human-machine interaction (HMI).