---

Wednesday, September 11

9:00-9:15 Welcome Address (HS1)

9:15-10:30 DAGM Award Session (HS1)

• German Pattern Recognition Award 2024
• DAGM MVTec Dissertation Award 2024
• Young Researchers' Forum Award 2024

10:30-11:30 PI Talk: Leif Kobbelt (HS1)

13:00-14:00 GCPR Oral Session 1: Learning Techniques (HS1)

13:00-14:00 VMV Oral Session 1: Rendering (HS2)

14:00-15:00 Keynote: Olga Sorkine-Hornung (HS1)

15:30-16:30 Research Spotlights (HS1)

Marina Evers: Uncertainty-aware Visual Analysis of Scientific Data
Abstract: Scientific data from different domains, such as physics, climate science, or medicine, is often associated with uncertainty. This uncertainty can, for example, originate in unknown parameters or initial conditions of the computational model or in random processes that underly the observed phenomenon. Incorporating this uncertainty into the analysis process is crucial when interpreting the data, deriving insights, and making informed decisions. In this talk, I will present my research on the visual exploration of scientific data. An interactive visual analysis provides insights into the influence of input parameters to results of numerical simulations and allows for understanding the structure of multi-dimensional parameter spaces. Another aspect covers the uncertainty-aware investigation of temporal evolutions including the correlation structure of spatial data. The applicability of the methods is shown based on examples from climate science and physics.

Paul Roetzer: Geometric Consistency in Shape Matching
Abstract: The shape matching problem refers to finding correspondences between two shapes. We consider 2D shapes (cyclic graphs where all vertices lie on a plane) and 3D shapes (triangles meshes in 3D space). Geometric consistency loosely defines neighbourhood preservation of correspondences between two shapes and is overlooked in many shape matching approaches as it leads to non-convex optimisation problems with a large set of constraints. Approaches which consider geometric consistency usually require specialised solvers and/or do make compromises in terms of solution quality/optimality. In this talk we learn how to use shortest paths to efficiently find geometrically consistent and globally optimal solutions for matching 2D shapes to 3D shapes and furthermore for matching two 3D shapes.

Anna Kukleva: Advancing Video Recognition with Less Supervision
Abstract: Deep learning has become an essential component of modern life, transforming various tasks across multiple domains such as entertainment, education, and autonomous driving. However, the increasing demand for data to train models for emerging tasks poses significant challenges, particularly in the video domain. Deep learning models heavily rely on high-quality labeled datasets, yet obtaining comprehensive supervision is resource-intensive and can introduce biases. Therefore, we explore strategies to mitigate the need for full supervision and reduce data acquisition costs for video understanding tasks. The first part of the discussion provides a comprehensive overview of my research in the video domain, focusing on unsupervised and weakly supervised methods and extending this knowledge to open-world settings. The second part of the presentation explores specific strategies for minimizing the need for precise annotations in multimodal video learning, addressing both large-scale pretraining and downstream tasks. Overall, this research contributes to a deeper understanding of data available for video tasks and its impact on learning dynamics, advancing methods that require less supervision.

16:30-17:30 GCPR Poster Session 1

16:30-18:10 VMV Oral Session 2: Information Visualization (HS2)

17:00-21:00 Industry Fair

---

Thursday, September 12

9:20-10:00 GCPR Oral Session 2: Medical Applications (HS1)

9:00-10:00 VMV Oral Session 3: Geometry (HS2)

10:30-10:35 Sponsor Spotlight: Zeiss (HS1)

10:35-11:30 Keynote: Torsten Möller (HS1)

13:00-14:00 GCPR Oral Session 3: 3D Vision (HS1)

13:00-14:00 VMV Oral Session 4: Scientific Visualization (HS2)

14:00-15:00 Keynote: Serge Belongie (HS1)

15:30-16:30 PI Talk: Christian Theobalt (HS1)

16:30-17:30 GCPR Poster Session 2:

18:00-21:00 Conference Dinner (Augustiner Stammhaus)

---

Friday, September 13

9:00-10:00 PI Talk: Tatiana von Landesberger (HS1)

10:30-11:30 Keynote: Cordelia Schmid (HS1)

13:00-14:00 GCPR Oral Session 4: Remote Sensing (HS1)

13:00-14:00 VMV Oral Session 5: Animation and Simulation (HS2)

14:00-15:00 PI Talk: Jörg Stückler (HS1)

15:00-15:30 Closing Session

• VMV / GCPR Best Paper Award
• Closing Remarks