Research Seminar Combinatorial Image Analysis

Contents

This seminar is about combinatorial optimization problems with applications in the field of image analysis. This includes graph decomposition problems for image segmentation and multiple object tracking, assignment problems for object matching, and unconstrained multi-linear optimization problems for inference in discrete graphical models. For analyzing and solving these problems, the seminar covers techniques from the areas of combinatorial optimization, computational complexity and algorithm engineering.

Organization

• Creditability: CMS-SEM, INF-AQUA
• Registration: The seminar is aimed primarily at doctoral students and is open to all qualified students. Students interested in taking this seminar are expected to send a topic proposal to mlcv@tu-dresden.de prior to the kickoff meeting and present their proposal during the kickoff meeting. Inspiration for topics can be found e.g. here.
• Kickoff meeting: APB-2026, Tuesday, 2023-10-10, 9:20-10:50
• Meetings: APB-2026, Tuesdays, 9:20-10:50
• Schedule:

  Date	Talk
  2023-10-10	Kickoff
  2023-10-17	Multicut game (FW, LH)
  2023-10-24	Min max correlation clustering (JI)
  2023-10-31	(public holiday)
  2023-11-07	Detection of levels of damage of concrete pavement (HGD)
  2023-11-14	?
  2023-11-21	Branch-and-cut algorithm for clique partitioning (MM)
  2023-11-28	Graph multi-separator problem (SZ)
  2023-12-05	Linear ordering problme (DS)
  2023-12-12	?Facets of lifted multicut polytopes (FN)
  2023-12-19	?Learning of partial Boolean functions (SB)
  2024-01-09	?Bregman optimization (HH)
  2024-01-16	Clustering of organoid images (JP)
  2024-01-23	An algorithm for a graph multi-separator problewm (JI)
  2024-01-30