Regensburg 2022 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

HL: Fachverband Halbleiterphysik

HL 30: Poster 2

HL 30.17: Poster

Donnerstag, 8. September 2022, 11:00–13:00, P3

object detection of patterned GaN using convolutional neural networks and synthetic data — •Mahdi Khalili Hezarjaribi, Uwe Rossow, Markus Etzkorn, Heiko Bremers, and Andreas Hangleiter — Institut für Angewandte Physik, Technische Universität Braunschweig

Employing a practical object detection algorithm, we have developed a process to detect GaN pyramid structures, extracted from SEM images. A procedure has been developed to generate synthetic images for training the algorithm instead of the drudgery of multiple imaging of samples. These synthetic data include noise, blurring, and other contributing factors in order to realize images that are accurate enough to train an object detection algorithm. A MobileNet algorithm has been employed for the Object detection process. The synthetic database proved pragmatic leading to a promising confidence value of 75% for detecting real objects.