Deep learning is a subfield of machine learning and is based on artificial neural networks with many processing layers. The technology enables computer systems to independently recognize complex patterns in large amounts of data.

Deep learning is particularly effective in the analysis of images, videos, speech, and sensor data. Unlike traditional algorithms, the relevant features do not need to be explicitly programmed but are automatically learned during training.

In industrial applications, deep learning enables the detection of surface defects, tool wear, material anomalies, or process deviations with high accuracy.

For quality control in automated laboratories, deep learning opens up new possibilities for the automated evaluation of samples and process conditions. By combining computer vision and deep learning, even complex quality characteristics can be analyzed reliably and reproducibly.

HERZOG uses deep learning methods within PrepMaster Analytics to automatically evaluate image data. Among other things, the algorithms support the detection of surface defects, tool wear, and other quality characteristics.