Institut für Informationsverarbeitung (TNT)

Timo Kaiser studied Mechatronics at Leibniz Universität Hannover. During his B. Sc. studies he focussed on Robotics. During his M. Sc. studies he focussed on digital image processing. In his master thesis "Multi-Objekt Tracking mittels eines Ende-zu-Ende Trainings" the multiple people tracking problem is modeled and optimized as a Conditional Random Field. He passed the Master's Examination in march 2020 and received his M. Sc. degree.

Since Mai 2020 he is working towards his Dr.-Ing. at the Institut für Informationsverarbeitung (TNT). His main research interests include multiple object tracking applied to time-lapse microscope sequences and uncertainty quantification.

During his jouney, he also worked on robust learning approaches, neural network pruning, outlier detection and interpretable machine learning. He works on the research project GreenAutoML4FAS, aiming to reduce the energy footprint of autonomous driving technologies, in conjunction with research groups from the LUH and the industry.

If you are interested in one of the above topics, feel free to contact him via the contact form.

Github page: https://github.com/TimoK93

Google Scholar: https://scholar.google.de/citations?user=L-SQX2YAAAAJ&hl=de

Publications

Show recent publications only

Conference Contributions
- Thomas Norrenbrock, Timo Kaiser, Sovan Biswas, Neslihan Kose, Ramesh Manuvinakurike, Bodo Rosenhahn
  CHiQPM: Calibrated Hierarchical Interpretable Image Classification
  The Thirty-ninth Annual Conference on Neural Information Processing Systems (NeurIPS), San Diego, USA, December 2025
  (pdfCode, Link) BibTeX
- Timo Kaiser, Thomas Norrenbrock, Bodo Rosenhahn
  UncertainSAM: Fast and Efficient Uncertainty Quantification of the Segment Anything Model
  Forty-second International Conference on Machine Learning (ICML), PLMR, Vancouver, July 2025
  (pdf, pdfGitHub, arXiv.org) BibTeX
- Thomas Norrenbrock, Timo Kaiser, Sovan Biswas, Ramesh Manuvinakurike, Bodo Rosenhahn
  QPM: Discrete Optimization for Globally Interpretable Image Classification
  The Thirteenth International Conference on Learning Representations (ICLR), April 2025
  (PDF, Link, GitHub) BibTeX
- Timo Kaiser, Ulman Vladimir, Bodo Rosenhahn,
  CHOTA: A Higher Order Accuracy Metric for Cell Tracking
  European Conference on Computer Vision Workshops (ECCVW), Springer, October 2024
  (pdfGitHub) BibTeX
- Patrick Glandorf*, Timo Kaiser*, Bodo Rosenhahn, (*contributed equally)
  HyperSparse Neural Networks: Shifting Exploration to Exploitation through Adaptive Regularization
  International Conference on Computer Vision Workshops (ICCVW), October 2023
  (pdf, pdfarXiv.org) BibTeX
- Timo Kaiser, Christoph Reinders, Bodo Rosenhahn,
  Compensation Learning in Semantic Segmentation
  Computer Vision and Pattern Recognition Workshops (CVPRW) , June 2023
  (pdf, pdfGitHub, arXiv.org) BibTeX
- Andrea Hornakova*, Timo Kaiser*, Michal Rolinek, Bodo Rosenhahn, Paul Swoboda, Roberto Henschel, (* equal contribution)
  Making Higher Order MOT Scalable: An Efficient Approximate Solver for Lifted Disjoint Paths
  International Conference on Computer Vision (ICCV), IEEE, October 2021
  (pdf, pdfarXiv.org) BibTeX
- Andrea Hornakova*, Timo Kaiser*, Bodo Rosenhahn, Paul Swoboda, Roberto Henschel, (* equal contribution)
  Higher Order Multiple Object Tracking for Crowded Scenes
  Computer Vision and Pattern Recognition Workshops (CVPRW) , June 2021
  (pdfPDF, GitHub, arXiv.org) BibTeX
Journals
- Timo Kaiser, Maximilian Schier, Bodo Rosenhahn,
  Cell Tracking according to Biological Needs - Strong Mitosis-aware Multi-Hypothesis Tracker with Aleatoric Uncertainty
  Transactions on Medical Imaging, IEEE, 2025
  (pdfIEEEexplore, DOI, GitHub, arXiv.org) BibTeX
Technical Report
- Timo Kaiser, Lukas Ehmann, Christoph Reinders, Bodo Rosenhahn,
  Blind Knowledge Distillation for Robust Image Classification
  arXiv, arXiv, November 2022
  (pdfarXiv.org) BibTeX