Research Seminar on AI: Learning Physical Displacement Fields of Particle Laden Fluid Flows using Deep Optical Flow Networks
- +49 241 80 90398
- E-Mail schreiben
Dienstag, 15.11.2022, 16.00 Uhr
A wide range of problems in applied physics and engineering involve learning physical displacement fields from data and so does Particle-Image Velocimetry (PIV). PIV is one of the key techniques in modern experimental fluid mechanics and is of crucial importance in diverse applications such as automotive, aerospace and biomedical engineering. The current state of the art in PIV data processing involves traditional handcrafted models that are subject to limitations including the substantial manual effort required, the limited spatial resolution, and difficulties in generalizing across conditions. Therefore, Christian Lagemann and his team explore new approaches for PIV processing leveraging a novel neural network architectures for optical flow estimation and introduced a new processing pipeline called RAFT-PIV. By contrast to the existing methods, the deep learning-based approach is general, largely automated and provides a spatial resolution which goes far beyond the currently used algorithms. Moreover, extensive experiments demonstrate that RAFT-PIV achieves state-of-the-art accuracy and generalization to new data, relative to both classical approaches and previously proposed optical flow learning schemes.
Currently, Christian is postdoctoral researcher in the Institute of Aerodynamics at RWTH Aachen where he also conducted his PhD studies. In his thesis, he introduced a novel deep neural network based approach for learning displacement fields in an end-to-end manner focusing on the specific case of Particle-Image Velocimetry. Before his PhD, he received a M.Sc. and B.Sc. in mechanical engineering at RWTH Aachen and was a visiting research scholar at City University of London. Now, he works on neural methods aiming at the discovery of underlying relations such as ordinary or partial differential equations solely from vision resp. observational data.