Konvexe Optimierung für Computer Vision

What it is about

You will learn the fundamentals of convex analysis and their application to optimization problems in image processing and computer vision. After the module you will be able to understand, apply and implement common first-order and proximal methods for typical CV tasks (e.g., image reconstruction, segmentation, matrix factorization).

What you will be able to do

• Master fundamental tools of convex analysis
• Apply first-order methods for convex problems with constraints and non-differentiable functions
• Derive saddle-point formulation and duality of energy minimization problems
• Understand and transfer convergence analysis of the Proximal Point Algorithm
• Independently solve convex optimization problems in computer vision numerically

What the module consists of

• VorlesungContents are presented and explained
• ÜbungTheoretical and practical exercises in groups for deeper understanding

Teaching method

• Vorlesung mit ÜbungLecture of the content; problem-solving groups work on exercises for deeper understanding
• Just-in-Time-Teaching (brief review and Q&A at the beginning of each lecture)Addressing comprehension issues and consolidating the material