Bachelor & Master Project
We offer projects for students in Media Informatics, Computer Science, Software Engineering, and Cognitive Systems. Projects can be completed individually or as a group.Topic Selection
At the start of each semester, we provide an updated list of available project topics here.Available Bachelor & Master Projects
Below is a list of currently available projects for students.
ControlNet for Data Augmentation in Semantic Segmentation for Electron Microscopy
Degree: Master Computer Science, Master Mediainformatics, Master Software Engeneering
Supervisor: Hannah Kniesel
Contact: hannah.kniesel@uni-ulm.de
Description: In this project, we will investigate the use of ControlNet for augmenting Semantic Segmentation datasets for Electron Microscopy. This includes investigating possible prompting schemas and conditions to maximize the ability of ControlNet to generate realistic EM data for training semantic segmentation models.
Further Reading:
Curating Bias in Datasets using Active Learning Inspired Uncertainty Measures
Degree: Bachelor Computer Science, Bachelor Mediainformatics, Bachelor Software Engeneering
Supervisor: Hannah Kniesel
Contact: hannah.kniesel@uni-ulm.de
Description: In this project, we will investigate the use of different uncertainty measures to curate bias in datasets without labels. Imagine training a segmentation model to segment different types of dogs, but there is very little training data on the corgi dog in the dataset. We want to use different active learning inspired uncertainty measures to find that the corgi is underrepresented in the training data.
Further Reading:
Curating Biased Datasets in Semantic Segmentation using AL Inspired ControlNet Guidance
Degree: Master Computer Science, Master Mediainformatics, Master Software Engeneering
Supervisor: Hannah Kniesel
Contact: hannah.kniesel@uni-ulm.de
Description: In this project, we will investigate the use of ControlNet Guidance (see paper in "further reading") for curating biased datasets. Imagine training a segmentation model to segment different types of dogs, but there is very little/no training data on the corgi dog in the dataset. We want to use ControlNet Guidance to generate more images which are representing the corgi dog in order to train on a fair, balanced dataset.
Further Reading: https://arxiv.org/pdf/2503.09221
Project Kick-Off
At the beginning of each semester, we hold a project kick-off meeting. Keep an eye out for our posters for dates and details.
- During the kick-off, we present current project topics and provide a brief introduction to relevant concepts.
- After the presentation, we offer a workshop-style session where you can:
- Test relevant methods.
- Discuss with other students to form project groups.
- Develop your own ideas by consulting with supervisors.
- During the meeting, you will need to register your project preferences by providing a first and second choice.
- Following the kick-off, you must contact the corresponing supervisor to arrange your first meeting.
Project Process
- Planning Phase: In this phase, the students familiarize themselves with the subject and collaboratively plans the project work with the supervisor. The outcome is a 1-2 page document (document template) outlining the work plan and the objectives of the thesis. When students work on projects in groups, you need to clearly mention their responsibilities for each project module.
- Exam Registration Phase: The proposal must be submitted no later than three weeks after the project kick-off. After submission, students must register for their respective exam module.
- Implementation Phase: During this phase, the objectives defined in the planning phase are realized. This is typically the most time-consuming phase of the project.
- Writing Phase: In the final phase, the student writes the project report in publication style (template) detailing the topic in relation to related work, and discussing both the implementation and the results. Publication style project reports are limited to a maximum of 8 pages, excluding references. Any additional information that exceeds this page limit should be included as Supplementary Material. This may include extended data tables, detailed methodology, additional figures or results, code, or any other content that supports the main text but is not essential to its core argument. The project can be written in either German or English (based on supervisor).
- Final Presentation: The final presentation (presentation template) must take place no later than one week after the teaching period (i.e., the first week of the semester break). Bachelor’s students are required to present for 10 minutes, followed by a 5-minute discussion (10+5), while Master’s students must present for 10 minutes with a 10-minute discussion (10+10). Possible dates are arranged in advance with the supervisor.
- Submission: The written project must be submitted to the supervisor no later than two weeks after the teaching period ends. Additionally, all code and data related to the project must be properly documented and shared with the supervisor.