Computer Vision I: Variational Methods
WS 2017/18, TU München
- 27.01.2018: Make sure you use the latest version of the lecture pdfs when preparing for the exam.
- 24.01.2018: Fixed typo in solution of sheet 8, uploaded solution of sheet 10.
Location: Room 02.09.023
Time and Date:
Wednesday, 10.15h - 11.45h
Thursday, 10.15h - 11.00h
Lecturer: Dr. Yvain Queau
The lectures are held in English.
Location: Room 02.09.023, also 02.05.014 for the practical part
Time and Date: Tuesday, 16:00h - 18:15h
Organization: Nikolaus Demmel, Christiane Sommer
Office hour: Thursday, 11:00h - 12:00h, or upon request
In addition to the exercise session on Tuesday, the computer room 02.05.014 is reserved for you Thursdays 11:00h - 12:00h, right after the lecture. You can of course use it most other times, when there is not a different tutorial (see here for more info). Also, Thursdays 11:00h - 12:00h, Christiane and Nikolaus will be available for questions in their offices, so feel free to come by. You can also come by another time, but it probably best to write an email before.
- Date: 22.02.2018
- Time: 10:30 - 12:30
- Place: MW 1801, Ernst-Schmidt-Hörsaal (5508.01.801)
- Registration: TUMonline (deadline: 15.01.2018)
There will also be a re-exam:
- Date: 26.03.2018
- Time: 10:30 - 12:30
- Place: Interims Hörsaal 2 (5620.01.102)
- Registration: TUMonline (deadline: 19.03.2018)
Only standard writing materials (=pens) are allowed in the exams. No cheat sheet, no additional paper, no book, no calculator. Enough paper will be provided.
Variational Methods are among the most classical techniques for optimization of cost functions in higher dimension. Many challenges in Computer Vision and in other domains of research can be formulated as variational methods. Examples include denoising, deblurring, image segmentation, tracking, optical flow estimation, depth estimation from stereo images or 3D reconstruction from multiple views.
In this class, I will introduce the basic concepts of variational methods, the Euler-Lagrange calculus and partial differential equations. I will discuss how respective computer vision and image analysis challenges can be cast as variational problems and how they can be efficiently solved. Towards the end of the class, I will discuss convex formulations and convex relaxations which allow to compute optimal or near-optimal solutions in the variational setting.
The requirements for the class are knowledge in basic mathematics, in particular multivariate analysis and linear algebra. Some prior knowledge on optimization is a plus but is not necessary.
A previous (very similar) version of this course was recorded in 2013. The videos can be found here.