TSBB15 Computer Vision

VT2022

This course ran the last time spring 2022. It has been replaced by TSBB33 and TSBB34.

Registration

If you intend to take the course but are not registered, make sure to register ASAP in Ladok, using the Student portal. You need to be registered on the course to receive course email, and to have results registered in Ladok. If you take the course but are not registered to any program at Linköping University, please contact the course examiner in order to make sure that you receive email about the course.

Course extent

This course is worth 12 ECTS credits, which corresponds to approximately 320h of work per student. The time is divided among the following activities:

More details can be found in Studieinfo (the web interface to the Bilda course database).

People

• Per-Erik Forssén, lectures, examiner
• Michael Felsberg, lectures

• Mårten Wadenbäck, lectures

• Johan Edstedt, computer lessons, computer exercises, project guide
• Karl Holmqvist, project guide
• Pavlo Melnyk, project guide

We all have our offices in the B-building, ground floor. Teacher offices are in corridor D (rooms 2D:513 to 2D:527), assistant offices are in Visionen. We recommend the use of email to request meetings.

Literature

In the course we will use the following literature:

1. Book: Richard Szeliski, Computer Vision: Algorithms and Applications, Springer Verlag (2nd ed. 2022). This book covers topics such as tracking, optical flow and image features. The book is available as an campus e-book via the LiU library. See also Book webpage.
2. Book: Klas Nordberg, Introduction to Representations and Estimation in Geometry (IREG). This compendium covers only the 3D geometry part of the course.
3. GIT: Other relevant literature, e.g. articles and masters theses can be fund in the TSBB15 GIT repository.

We have collected pointers to relevant literature in the column “Material” in both of the “Lecture and Lab schedules” below. These either refer to one of the two books above, or to an article in the TSBB15 GIT repository. There are also additional pointers that may be of use, e.g. to introductory videos and Wikipedia pages.

Examination

The course has a written examination and a mid-term exam. These are offered at the following occasions during 2022 (see Tentabokning):

Exam related questions may be asked in the Exam Q and A chat in the course teams group]. We will answer when we have time. Make sure to tag a teacher to get a response.

Old exams

See the page on previous exams for a selection of old exams given in the course.

How to use the old exams

We recommend that you use these exams as pointers into the literature (“instuderingsfrågor”). To answer the exam, read corresponding discussions in your notes, the course book, and the lecture slides. While it is faster to look at other people’s exam answers, this merely gives you answers, while reading and thinking also results in understanding.

Grades

• Details on grading can be found on the page for grading criteria.

Lecture and lab schedule

Lecture and lab schedule VT2

Projects

The two projects are conducted in groups of 5, 4 or 3 students (in order of preference). We aim for 3 project groups this year. Assignment into groups is made on the introductory lecture for project 1.

List of project groups VT2022

• Project 1: Object Tracking
• Project 2: 3D Reconstruction

General resources

We suggest and allow you to use the following software:

• OpenCV (Open Source Computer Vision).

  Version 3.4.3 is installed on the department’s Linux computers (first you need to issue the module add prog/opencv/3.4.3 command). See also the minimal example OpenCV progam (contributed by Gustav Häger).

  Read the section on OpenCV in the hints and pitfalls page. There is a cheat sheet for OpenCV. An important exception is the Background modelling with mixtures of Gaussians, which you are NOT allowed to use (as you’re supposed to learn this in the course).

• Python with OpenCV bindings.

  Python 3.6.9 is installed in the computer labs. Essential libraries such as Matplotlib, NumPy, and SciPy are also available.

• VLFeat has a a useful code library, both for Matlab and C/C++. For example, it has an alternative implementation of SIFT, and also an implementation of MSER. Both are made by Andrea Vedaldi.
• The Visual Geometry Group at Oxford University maintains code for affine invariant region detectors, produced in cooperation with other groups.
• The Computer Vision Laboratory at ETH provides an implementation of SURF.
• An IDE, e.g. PyCharm community edition, which is installed in the computer labs after loading the right module (see module avail) (remember to say no to creation of a shortcut on /usr/local when you start pycharm).

Project repositories

Project code should be developed under versioning control, with changes tracked according to LiU-ID of the participating group members.

• Project groups should create their repositories in the LiU Git. Note: this is not GitHub, and GitHub should not be used.
• Project guides and examiner should be given “reporter” access to the group repositories.

Hints and pitfalls

Based on experience from previous year’s projects, we have accumulated a list of hints and pitfalls for the projects. Read them carefully before starting your project work.