Wintersemester

Computer Vision (VU)

Course Details

Lecturer

Robert Sablatnig
Marco Peer

LVA-Nr.

183.585

Typ

Vorlesung mit Übung (VU)

Links

TISS

TUWEL

Dates

  • Computer Vision (VU) is planned to be in presence in WS2022/2023. The exercise part will mainly take place virtually.
  • A timetable for orientation of the lecture part is provided here.
  • Exercise part: Tutoring hours TH 9:00 – 11:00 as well as 11:00-13:00 in a 2-week rhythm, see the schedule in TUWEL (starts on 3.10.2022).

Location

  • Lecture part: Monday, 15:00 – 19:00 and Tuesday, 17:00 – 19:00 in FAV Hörsaal 1 – INF.
  • Exercise part – tutor hours: virtual via Zoom ( 1 registration per group via TUWEL required)
  • Two submission talks via Zoom

The exercise part will be handled via the associated TUWEL course (This will be visible for students at the beginning of October). In the course of the semester, various tasks are to be solved for the exercise part in groups of 3 using Python. For support, there will be supervised practice times, which will take place 6x in the semester (approximately every 2 weeks). Please register in TUWEL for a group of 3. Attendance is compulsory only for the 1st session on 03.10.2022.

Lecture exam: Registration in TISS

 


Computers are still limited in their ability to interact with the world and with human users because they cannot “see.” When dealing with computer vision, we must also try to understand the physics of the world, for example, how light reflects off surfaces, how objects move, and how all of this information is projected onto an image through the optics of the camera. It also requires to develop algorithms that reconstruct some of these physical properties from one or more images. This “inverse” problem is actually a big puzzle. Information is lost when the three-dimensional world is projected onto a two-dimensional image, so how do we recover this information from an image? This course shows the basic algorithms that make this goal achievable and develops methods for solving various inverse problems. But vision is more than just reconstructing the 3D world from 2D images, it also deals with the extraction of semantics. The course also demonstrates basic machine learning and probabilistic inference methods to solve this problem.

Even if you do not plan to study computer vision further, the basic tools and techniques used there can be useful in many other areas. For all those who want to learn more about computer vision, this course provides a good basis for further study.

Content:

  • A perspective on many areas of current computer vision research
  • An implementation of programming tasks to gain practical experience with working with images and image sequences
  • The application of linear algebra and calculus for something you can use in the real world

Organization: