In her recent research, Julia Hartung has been applying Machine learning (ML) technology to smart manufacturing where it can be harnessed to provide insights into complex processes without requiring deep domain expertise.
In a paper released in 2022, Hartung and her colleagues used deep learning algorithms to determine a 3D reconstruction from a single 2D grayscale image. 3D reconstruction can be used for quality control of laser-welded components because the height values contain relevant information that is not visible in 2D data.
In this recent project, the researchers compared three deep learning methods, namely stacked autoencoder (SAE), generative adversarial networks (GANs) and U-Nets to see how they compared when reconstructing a 3D image from a 2D grayscale image of laser-welded components.
You can find Hartung online at linkedin.com/in/julia-hartung-523a5416b/ or researchgate.net/profile/Julia-Hartung-2.
Organisation: Trumpf Laser
Role: currently pursuing a doctorate degree with the Karlsruhe Institute of Technology
Based in: Karlsruhe, Germany
Education: Master's in applied informatics, Hochschule Esslingen, University of Applied Sciences