Hannah Kniesel

I am a PhD student at Ulm University where I am supervised by Timo Ropinski and Pedro Hermosilla.

In my ongoing research, I explore deep learning techniques, emphasizing their relevance in domains marked by distinctive data challenges. The focal point is introducing deep learning to fields with inherent complexities such as small datasets, non-object-centric data, and pervasive noise. Beyond this, my research is dedicated to crafting models distinguished by their innate adaptability and retrainability, crucial for seamlessly accommodating shifts in the dataset. This adaptability is pivotal for the effective integration of deep learning into dynamic research domains. By addressing these issues, my research seeks to open new avenues for the integration of deep learning in diverse fields, like biomedicine, where traditional approaches may fall short.

I completed my master's degree in computer science at the Ulm University. In my master thesis I worked on the improvement of electron tomography (ET) by integrating neural networks similar to NeRF into the reconstruction process. Additionally, I introduced a noise module that was able to disentange signal from noise, leading to a noise free reconstruction from noisy projections. My master thesis was supervised by Timo Ropinski and Pedro Hermosilla. Additionally, I worked as a research assistant in the Visual Computing Group. Before that, I pursued my bachelor's degree in media informatics at Ulm University. In my bachelor thesis I developed an efficiant way of visualizing and rendering amyloid beta fibrils. I also gained practical experience through a student trainee at Daimler, with focus on the research and development of dialogue systems.

Email  |  GitHub  |  Google Scholar  |  Twitter

Hannah Kniesel

I am a PhD student at Ulm University where I am supervised by Timo Ropinski and Pedro Hermosilla.

In my ongoing research, I explore deep learning techniques, emphasizing their relevance in domains marked by distinctive data challenges. The focal point is introducing deep learning to fields with inherent complexities such as small datasets, non-object-centric data, and pervasive noise. Beyond this, my research is dedicated to crafting models distinguished by their innate adaptability and retrainability, crucial for seamlessly accommodating shifts in the dataset. This adaptability is pivotal for the effective integration of deep learning into dynamic research domains. By addressing these issues, my research seeks to open new avenues for the integration of deep learning in diverse fields, like biomedicine, where traditional approaches may fall short.

I completed my master's degree in computer science at the Ulm University. In my master thesis I worked on the improvement of electron tomography (ET) by integrating neural networks similar to NeRF into the reconstruction process. Additionally, I introduced a noise module that was able to disentange signal from noise, leading to a noise free reconstruction from noisy projections. My master thesis was supervised by Timo Ropinski and Pedro Hermosilla. Additionally, I worked as a research assistant in the Visual Computing Group. Before that, I pursued my bachelor's degree in media informatics at Ulm University. In my bachelor thesis I developed an efficiant way of visualizing and rendering amyloid beta fibrils. I also gained practical experience through a student trainee at Daimler, with focus on the research and development of dialogue systems.

Email  |  GitHub  |  Google Scholar  |  Twitter

News
  • 01/2024 Our paper Weakly Supervised Virus Capsid Detection with Image-Level Annotations in Electron Microscopy Images has been accepted at ICLR!
  • 07/2023 Our paper to my master's thesis Clean Implicit 3D Structure from Noisy 2D STEM Images has been accepted at CVPR!
  • 11/2022 My master thesis on the improved tomographic reconstruction was honored with the Harald Rose Award for imaging and analytical procedures.
  • 11/2022 I was honored with the eXXcellence Award for the best master's thesis in the field of computer science.
  • 04/2022 Start of my PhD journey at Ulm University.
  • 02/2022 I successfully defended my master's thesis on improved electron tomographic reconstruction using deep learning.
  • 09/2020 Our follow up paper Real-Time Visualization of 3D Amyloid-Beta Fibrils from 2D Cryo-EM Density Maps to my bachelor's thesis was accepted at EG VCBM
  • 2019 I successfully defended my bachelor's thesis on the efficient visualization of amyloid beta fibrils.
Research
Evaluating Text-to-Image Synthesis: Survey and Taxonomy of Image Quality Metrics
Sebastian Hartwig, Dominik Engel, Leon Sick, Hannah Kniesel, Tristan Payer, Poonam Poonam, Micheal Glöckler, Alex Bäuerle, Timo Ropinski,
Preprint, 2024
project page | bibtex
Weakly Supervised Virus Capsid Detection with Image-Level Annotations in Electron Microscopy Images
Hannah Kniesel, Leon Sick, Tristan Payer, Tim Bergner, Kavitha Shaga Devan, Clarissa Read, Paul Walther, Timo Ropinski, Pedro Hermosilla
International Conference on Learning Representations (ICLR), 2024
project page | bibtex
Künstliche Intelligenz in der Radiologie – jenseits der Black-Box
Luisa Gallee, Hannah Kniesel, Michael Götz, Timo Ropinski
RöFo - Fortschritte auf dem Gebiet der Röntgenstrahlen und der bildgebenden Verfahren, 2023
project page | bibtex
Clean Implicit 3D Structure from Noisy 2D STEM Images
Hannah Kniesel, Timo Ropinski Tim Bergner, Kavitha Shaga Devan, Clarissa Read, Paul Walther, Tobias Ritschel, Pedro Hermosilla
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2022
project page | bibtex
Real-Time Visualization of 3D Amyloid-Beta Fibrils from 2D Cryo-EM Density Maps
Hannah Kniesel, Timo Ropinski, Pedro Hermosilla
Eurographics Workshop on Visual Computing for Biology and Medicine (EG VCBM), 2020
project page | bibtex
Teaching

2023 - today

Lecture: Digital Media

I currently accompany the lecture "Digital Media" and "Basics of Media Informatics" for bachelor (undergrad) students held by Timo Ropinski at Ulm University. I supervise and create exercises about different types of media, as well as their encoding and representation. We also include basics of deep neural networks in the lecture.

2022 - today

Projects and Theses

I offer the supervision and mentoring of bachelor (undergrad) and master (grad) students in regular projects/theses. Ideas are designed and formed in cooperation with the students. I accompany, support and teach the students during the implementation and finally guide through the documentation and writing phase.

2022 - today

Seminar "Computer Vision"

I design, implement and accompany the seminar for bachelor (undergrad) students. During this seminar, we give a short introduction into current reseach in computer vision, dominated by deep learning. Students are then guided through the process of reading topic specific academic papers and related work, writing their own academic paper and finally mimicing the review process of conference submissions.


template adapted from this and this awesome websites!