Open Student Projects

How To Apply

Please read this section carefully. I might not answer you if you do not follow the instructions and I am very busy at the moment (or just reply with a link here).

In order to apply write an email to zorah.laehner@uni-siegen.de with at least the following information:

Optional but appreciated:

I will then send you a small programming task related to 3D geometry processing (no prior knowledge required) which you have a week to solve. Afterwards we have a meeting where we will discuss your solution, I will ask some questions about Bachelor level math or data structure topics to make sure you have the needed background knowledge, and I give you a more detailed overview over the project(s).

Visual Computing Thesis Course

If you are not specifically interested in writing a thesis with me but just in the general topic of visual computing, consider our joint training program with the visual computing chair of Prof. Keuper: Moodle Course Mattermost Channel This is a 2-4 weeks course that gives an overview of the skills necessary for the successful completion of a Studienarbeit/Projektarbeit/Thesis. Each student that successfully completes this course is guaranteed a thesis topic for that semester. If you are not set on working with me, this is the preferable and safer way to obtain a thesis topic in the visual computing area. This program is also a good opportunity to self-review and to see if you have the required knowledge and understanding of some basic implementations in computer vision and image processing. Students that do not succeed will be recommended courses they can study to learn some basic principles they may be lacking before they re-apply.

Available Projects

Currently available spots: 0

(there are more topics here than available spots but I can only supervise a certain amount of students at the same time, please do not write emails if it says 0 spots available)

All projects related to deep learning need to be implemented in PyTorch.

Survey of Autoencoder Architectures

Summary: Survey, implement and compare existing autoencoder architectures for point clouds, voxel grids and triangle meshes.

Suitable for: Bachelor Thesis, Master Thesis, Studienarbeit

Requirements: Python, passed our Deep Learning lecture (exception for Bachelor students)

Analysis of Functional Maps framework using different basis sets

Summary: Survey, implement and compare the behavior of functional maps when replacing the default Laplace-Beltrami eigenbasis with different sets of basis functions.

Suitable for: Bachelor Thesis, Master Thesis, Studienarbeit

Requirements: Matlab or Python experience

Related Reading: Functional Maps: A Flexible Representation of Maps Between Shapes, Ovsjanikov et al., 2012