Lab course projects

Master studies, 6 ECTS

We offer a wide variety of possible projects in computer vision, vision related deep learning, and biomedical image analysis. We stopped listing particular projects here, since they get outdated quickly. We rather work out an up-to-date project on request. If you are sufficiently certain that you want to do a project in our group, please contact Prof. Thomas Brox. Do not forget to mention which relevant courses you have taken. Since the 6 ECTS are not sufficient to start a reasonable research project, we usually ask for the commitment to extend the lab course later with a Master project. The lab course must be finished within a single semester.

In the winter semester we usually offer as an alternative a GPU programming course with a final (GPU related) project.

Information on project organization, learning goals, hardware usage.

Running projects

Finished projects

• Egomotion from optical flow(Manuel Bühler, 2016)
• Cell segmentation and tracking using the U-net and Conservation Tracking (Jan Sosulski, Florian Krämer, 2015)
• Fully automatic calibration of multiple cameras to a single world coordinate system with bundle adjustment (Markus Frey, 2015)
• Determining motion boundaries from image boundaries using deep descriptor matching (Maxim Tatarchenko, 2014)
• Comparison of two methods that generate object region hypotheses (Tobias Wieland, 2013)
• Matlab volume renderer (Raphael Schmitt, 2012)
• Fast stereo with semi-global matching on the GPU (Martin Senk, 2011)
• Efficient 3D rotation invariant features for detection on the GPU (Daniel Kuhner, 2011)
• Identifying spatially consistent point correspondences by finding maximal cliques in graphs (Thomas Rinklin, 2011)
• Hierarchical diffusion-based interpolation on the GPU (Alejandro Alfaro, 2011)
• 3-layer latent-SVM (Tatiana Jiménez Cárdenas, 2011)