TSBB33 3D Computer Vision

This course covers the algorithms and estimation problems used to infer 3D structure from images. The course covers both the mathematics used, and how these are put into practice in algorithm implementation. The course features two computer labs, and a 3hp project where groups of participants together will implement a structure from motion system. The project is presented both in written form and orally at a seminar. The course ends with a written exam on the theoretical content.

New course! This course will run for the first time in spring 2023. It is based on the geometry and multi-view stereo parts of the course TSBB15 Computer Vision.

3D Reconstruction Flowchart

General Information

Lecture Schedule VT2023

Below is an up-to-date list of all staffed lectures and computer labs.
Additional, unstaffed computer lab time can be found in TimeEdit.

All lectures are given in Systemet and all labs are in Olympen, both located at the Department of Electrical Engineering.

Date,Time,Room Activity Teacher Literature
March 27: 10.15-12
Lecture 1
Introduction, 3D perception, local features
Per-Erik Forssén CVAA: 7.1
Background geometry: IREG ch. 7-9,13
March 28: 13.15-15
Lecture 2
Maximum likelihood, RANSAC, the essential matrix
Mårten Wadenbäck IREG: 10.5 (E), 15.4 (PnP), 16.3 (E est.)
March 30: 08.15-10
Lecture 3
Non-linear least squares estimation, and robust error norms
Per-Erik Forssén K. Madsen, NL Least Squares Tutorial
Z. Zhang, Parameter Est. Techniques, Ch. 9.
April 11: 13.15-17.00
Non-linear optimization and gold standard
Johan Edstedt
Ioannis Athanasiadis
CE1 Lab sheet
April 13: 08.15-10
Lecture 4
Multiview stereo, correspondence fields, and triangulation
Per-Erik Forssén Furukawa MVS-tutorial, PatchMatch paper
IREG: 10.4, CVAA: 12.1, 12.3
Jump Flooding paper
April 17: 10.15-12
Lecture 5
Representation of 3D rotations
Mårten Wadenbäck IREG: ch. 11
April 24: 10.15-12
Lecture 6
Absolute and relative camera pose, and minimal cases
Mårten Wadenbäck CVAA: 8.1.5, 11.2
IREG: ch. 15
April 25: 17.15-21
Densification with PatchMatch
Johan Edstedt
Ioannis Athanasiadis
CE2 Lab sheet
Skeleton code
Cheatsheet notebook
April 27: 8.15-10
Lecture 7
Structure from motion, bundle adjustment, and project start
Per-Erik Forssén IREG: 15.1,15.12 (3D alignment), 21 (SfM)
CVAA: 8.1.5, 8.3.1 (3D alignment), 11 (SfM)
May 8: 10.15-11
Guest Lecture 1
Martin Svensson
May 15: 10.15-11
Guest Lecture 2
Maxar Sweden
Leif Haglund
May 23: 13.15-17
Final Seminar
Project presentations, exam discussion, and wrap up
Per-Erik Forssén

3D reconstruction project

Half the course consists of a 3D reconstruction project. The project starts on April 27, and ends with a final seminar on May 23. The project builds on top of the two computer exercises.

For more information see the project description page.