Computer Vision and Machine Learning SS'25
Vorlesung mit Übung

Current information

The live sessions for the lecture have concluded. If you have any questions please send them to cvml@cg.cs.tu-bs.de.

Exam Date: 26.08.2025, 16:00-17:30, Audimax. Details here.

Official registration for the exam is open! Exam Date: 26.08.2025, 16:00-17:30, Audimax. Details here.

Please note that the exam for the Summer Semester will take place on the 26.08.2025 as decided by the the Prüfungsamt, details here.

You need to list all members of your exercise team (link for registering provided under the pdf for the exercise's kickoff slides further down) before the 22.4.2025 at 23:59. People not registered till then will not be able to take part on the exercises.

The lecture has reached maximum capacity, thus the enrollment is closed (both for lecture and exercises).

You can find an example exam here. Please note that this is shorter and it was handed out during the Corona pandemic. Therefore it is not entirely representative, but hopefully gives you at least some idea what to expect.

Language

Since the course is also offered in the Master Data Science program, the language of the course is English.

Description

After successful completion of this module, students have a basic understanding of the development of complex computer vision applications. They are able to understand computer vision problems and to design and effectively implement suitable (AI-based) solutions.

Registration

To participate in the lecture and exercise, you can register on our website under "Teaching -> Course Enrollment" (direct link https://www.cg.cs.tu-bs.de/teaching/students ).

Studip is not used for this course.

Registration for the exam is done through the Examination Office. 

Bachelor/Master Note

Upon request (form available from the Examination Office), this course can usually also be taken in the Bachelor's program. Please refer to the module handbook and your examination regulations to see if this course is also regularly offered in your Bachelor's program.

Content

The content may change until the start of the lecture.

- Image Acquisition
- Image Processing Basics
- Deep Learning
- Feature Detectors and Descriptors
- Dense Correspondences / Optical Flow
- Parametric Interpolation
- Epipolar Geometry
- Stereo and Multi-View Reconstruction
- Camera Calibration
- Video Matching
- Morphing and View Interpolation
- Neural Radiance Fields
- Object Detection
- Motion Capture
- Machine Learning for Computer Vision Problems
- Computer Vision for Special Effects

Location and Time

Lectures

The lecture is conducted as Inverted Classroom, i.e. you will have to watch the provided video/material in advance and the lecture time can be used for questions and in-depth subjects.

The material for the next session, as well as the lecture slides, are usually provided here one week in advance.

The password will be given in the lecture and can be requested at cvml@cg.cs.tu-bs.de if necessary.

LIVE sessions are held regularly every Tuesday, except during field trip week. Any updates will be published in the agenda.

Agenda

Will updated soon, stay tuned.

15.04.2025 Introduction and Image Acquisition [pdf1][pdf2][video] (Image Acquisition can be skipped if you already attended the module "Digital Image Processing" last semester)

22.04.2025 LIVE Session [pdf]

29.04.2025 Digital Image Processing Basics [pdf1][pdf2][video1][video2] (can be skipped if you already attended the module "Digital Image Processing" last semester)

06.05.2025 LIVE Session [pdf]

13.05.2025 Machine Learning Basics [pdf][video] (can be skipped if you already attended the module "Digital Image Processing" last semester)

20.05.2023 LIVE Session [pdf]

27.05.2023 Features [pdf][video]

03.06.2025 Optical Flow [pdf][video]

09.06. - 15.06.2025 Excursion week (no lecture or exercise)

17.06.2025 Parametric Transformations and Scattered Data Interpolation [pdf][video]

24.06.2025 Epipolar Geometry and Stereo [pdf][video]

01.07.2025 Video Matching, Morphing, and View Synthesis [pdf][video]

08.07.2025 Structure from motion [pdf][video]

15.07.2025 Neural Radiance Fields [pdf][video]

18.07.2023 LIVE Session / Q&A Session [pdf]

22.08.2023 LIVE Session / Q&A Session / Exam preparation

Exercises

In the exercises, programming will be done in Python with OpenCV and PyTorch.

The exercise tasks will be uploaded Thursdays and discussed on the following Thursday in the exercise session.

The tasks of each exercise sheet must be completed in groups of four to five people and uploaded to the Git repository by Wednesday 23:59 at the latest. Don't forget to include names and matriculation numbers in the repository.

The practical tasks must be demonstrated in the exercise session. Working groups of up to five people are allowed, but everyone in the group must be able to answer any questions about the tasks and the code independently.

Timeslots and order of teams for presentation (grey are teams without submission)

The frameworks and solutions have been tested on the computers in the CIP Pool. Unfortunately, we cannot guarantee direct support for other systems.

A computer with Linux or Windows is required for completing the tasks. The functionality of the framework under Mac OS/X cannot be guaranteed. If you encounter any issues, please contact us via email at cvml@cg.cs.tu-bs.de.

Session plan

Exam

The examination date can be found under Location and Time.
Any changes will be announced in the lecture and on this website in time.

• Exam format: written exam (90 minutes)
• Remember to arrive in time, we will open the doors 15 minutes before the starting time.
• Certificate acquisition by passing the exam (at least 50% of the points)
• Requirement for module completion: at least 50% of the points achieved in the exercises.
• Exam participation is recommended for every course of study!
• Students must register with the examination office!

Exam details

Below you can see the grading scale of the exam from March 30, 2021.

The exam results were sent to the participants by email.

You have the opportunity to raise an objection until April 8, 2021. Please send an email to cvml@cg.cs.tu-bs.de.

A meeting for clarification will be offered on April 9, 2021, at 1 pm if necessary (so far, there has been no corresponding request).

A sample solution can be found here.

Trial exam

A small trial exam can be found here.

Repeat Exam

If you have failed the regular exam or wish to claim a free attempt, you have the opportunity to take the repeat exam.

If that would be your last chance (usually the third) to pass the course, then please continue reading the section "Supplementary Exam"!

In semesters where the course is taught this is simply the regular exam. So all you need to do is register for the regular exam.

In semesters, where the course is not taught, it usually takes place during the first lecture-free week. To register, please sign up for the exam regularly at the examination office and send a corresponding email requesting a repeat exam to sekretariat@cg.cs.tu-bs.de, providing your full name, matriculation number, TU email address, number of attempt, and degree program.

The repeat exams are usually held collectively. A date will be communicated to you in due time after registration.

Please note that besides the exam you need to have successfully participated in the exercises to successfully pass the module. The exercises are only offered when the course is held!

Supplementary Exam

If you need an oral supplementary exam because you have not passed the exam in your final regular attempt, please send an email requesting an oral supplementary exam to sekretariat@cg.cs.tu-bs.de immediately after receiving the grade (i.e., when it has been officially registered with the examination office), providing your full name, matriculation number, TU email address, and degree program.

The oral supplementary exams are usually held collectively once per semester. A date will be communicated to you in due time beforehand.

Requirements

• Programming skills, preferably in Python

Literature

• Richard J. Radke: Computer Vision for Visual Effects, Cambridge University Press
  
• Richard Szeliski: Computer Vision: Algorithms and Applications, Springer Verlag
• D. Forsyth and J. Ponce: Computer Vision: A Modern Approach. Prentice Hall
• Goodfellow et al.: Deep Learning. Das umfassende Handbuch, MIT-Press