Johannes Künzel is a Research Associate at the Computer Vision & Graphics group in the Vision and Imaging Technologies Department.
He joined the group as a Research Associate in 2016. His research interests are camera pose estimation, the registration of high-definition 3D models, and the strengthening of computer vision algorithms against the influence of challenging weather conditions.
He received a B.Sc. degree in 2014 and an M.Sc. degree in 2016 from Ilmenau Technical University.
Publications
Articles in Journals
2020
- Jan Waschnewski, Ralf Hilpert, Daniel Sauter, Peter Eisert, Johannes Wolf Künzel, Birgit Schalter, Klaus Sympher, Ulrich Jöckel, Klemens Kresin, Karl-Heinz Franke, Daniel Kapusi, Rene Döhring, Philipp Woock, Florian Zimmermann
Sewer Status Detection: Perspectives with Innovative 3D Image Data and with Artificial Intelligence in 2D and 3D Image Analysis Based on the Example of the BMBF Project AUZUKA,
KA Korrespondenz Abwasser, International Special Edition, pp. 6-15, August 2020
2019
- Daniel Kapusi, Daniel Kapusi, Karl-Heinz Franke, Peter Eisert, Johannes Wolf Künzel, Florian Zimmermann, Jan Waschnewski
In den Tiefen Berlins,
inspect, WILEY-VCH Verlag, vol. 1, pp. 48–50, January 2019
Conference Contributions
2024
- Johannes Wolf Künzel, Anna Hilsmann, Peter Eisert
BTSeg: Barlow Twins Regularization for Domain Adaptation in Semantic Segmentation,
DAGM GCPR 2024, Munich, Springer Lecture Notes in Computer Science , September 2024, doi: https://doi.org/10.48550/arXiv.2308.16819, arXiv: https://arxiv.org/abs/2308.16819
2023
- Johannes Wolf Künzel, Darco Vehar, Rico Nestler, Karl-Heinz Franke, Anna Hilsmann, Peter Eisert
System for 3D Acquisition and 3D Reconstruction Using Structured Light for Sewer Line Inspection,
Proc. 18th Int. Joint Conf. on Computer Vision, Imaging and Computer Graphics Theory and Applications, Lissabon, ISBN: 978-989-758-634-7, February 2023, doi: 10.5220/0011779900003417, arXiv: https://arxiv.org/abs/2303.02978
2022
- Clemens Peter Seibold, Johannes Wolf Künzel, Anna Hilsmann, Peter Eisert
From Explanations to Segmentation: Using Explainable AI for Image Segmentation,
International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, virtual, February 2022, arXiv: http://arxiv.org/abs/2202.00315
2018
- Johannes Wolf Künzel, Thomas Werner, Ronja Möller, Peter Eisert, Jan Waschnewski, Ralf Hilpert
Automatic Analysis of Sewer Pipes Based on Unrolled Monocular Fisheye Images,
IEEE, Lake Tahoe, USA, March 2018, doi: 10.1109/WACV.2018.00223, Winter Conference on Applications of Computer Vision (WACV)
Projects
BerDiBa
“Berliner Digitaler Bahnbetrieb”
The automated rail transport is one essential cornerstone of a modern and sustainable transportation system. In order to achieve this goal, BerDiBa drives the progress in three major aspects: self-driving trains, automated remote control, and predictive maintenance. In collaboration with TU Berlin, we focus on the last aspect, which provides challenging research opportunities, like the registration between images taken in different weather conditions and/ or different seasons. This leads to the severe violation of fundamental assumptions like illumination consistency. Therefore, we research new methods to bridge this domain gap and to enable the robust deployment of computer vision on the rails of tomorrow. Website
GEMIMEG II
Secure and robust calibrated measurement systems for the digital transformation.
BMBF Project Auzuka
The inspection of sewer pipes is mandatory to guarantee their functionality. At present, mobile robots with cameras are used to tackle this task. The manual process of damage detection and classification is error prone because of the repeating and tiresome work. Therefor the goal of this project is the development of a system, which is capable of assisting the employee with the automatic detection and classification of damages.