|Date:||02.12.2021 / 16-17 Uhr|
|Speaker:||M.Sc. Bastian Lenz |
Research Fellow in the Section Surface Engineering at Leibniz-IWT
|Title:||Application of deep learning image recognition techniques for characterization of thin coatings|
|Registration:||see form at the bottom of the page |
Registration deadline: 02.12.2021
Characterization methods are an important part of surface engineering and coating development. By using deep learning image classification and object detection techniques, some of them can be automatized to meet the latest digitalization demands.
The Rockwell-C and the Scratch-test are established methods for coating adhesion determination. Since they are based on the classification of light microscopic images, it is possible to substitute humans with trained special neuronal networks, the convolutional neural networks. These so-called CNN's are already being used successfully in a variety of other applications, and can be adapted for material science purposes.
As an extension of Vickers hardness testing, the Palmqvist method utilises crack formation at the edges of Vickers indentations to evaluate the fracture toughness of bulk materials and coatings. By using CNN-based object detection, the necessary indentation and crack size measurement can be automatized.