392226 Project: Probabilistic Visual Computing (Pj) (SoSe 2025)

In this MSc project, you will gain practical experience with Probabilistic Computer Vision and Computer Graphics methods. Modern approaches typically contain a deep learning component and geometric representations of the objects that are reconstructed or rendered. Possible directions are

• testing and comparing existing methods in a new application scenario,
• extending them to work better/faster/robustly, and
• combining existing methods to get the best from both worlds.

A particular focus is on domains where uncertainty is important and on how to model this uncertainty probabilistically. Concrete examples are:

• Human pose & shape estimation
• Uncertainty estimation
• Bias reduction or privacy
• NeRF, Gaussian Splatting, and other 3D reconstruction techniques
• Unsupervised representation learning

The following slides provide additional details and context to the existing visual computing modules at Bielefeld University:
https://docs.google.com/presentation/d/e/2PACX-1vRGyE0j_cU324qp-vOEE4hEBPn3HdwliZQqeg_vT0w7Yt9EUXmgFB6NamL_ynqeBMYRgD-cdvfvXobA/pub?start=false&loop=false&delayms=3000

Requirements for participation, required level

Required skills: Python (>1 year), Machine learning, and deep learning basics.
Optional skills: PyTorch or TensorFlow experience, computer graphics, computer vision (lectures or projects).

Students don't need to be experts in all aspects of visual computing but have a strong background in one of graphics, vision, and machine learning. These skills are typically acquired by lectures, private projects, hackathons, and coding competitions.

Bibliography

The projects will center around existing publications from the CVPR and SIGGRAPH conferences or other respected computer vision and graphics venues.

External comments page

https://docs.google.com/presentation/d/e/2PACX-1vRGyE0j_cU324qp-vOEE4hEBPn3HdwliZQqeg_vT0w7Yt9EUXmgFB6NamL_ynqeBMYRgD-cdvfvXobA/pub?start=false&loop=false&delayms=3000