Tentalabbet

TSBB33 · 6 credits · Advanced level (A1N) · Datateknik, Elektroteknik

3D Computer Vision

Active · next start VT 2027

64%

passed the latest ordinary exam

TEN1 · 2 Jun 2026 · 14 students

59%

passed the ordinary exams, all sittings over the last 5 years

TEN1 · 41 students

Part of 8 programmes

Exam

TEN1 Written Examination

  • 5
  • 4
  • 3
  • U

Ordinary exams

59% passed · 41 students

  • 64%14
  • 38%13
  • 71%14

Re-exams

93% passed · 14 students

  • 83%6
  • 100%8
  • Pass the re-exam: 93% (14 students), compared with 59% (41 students) on the ordinary exam
  • Get 4 or 5: 24% (55 students)
Show small sittings (5)

Sittings with fewer than 5 students, often credit transfers or special sittings. They are not part of any figure on this page.

  • 100%2
  • 0%1
  • 0%1
  • 67%3
  • 33%3

Other modules

Partial exams, hand-ins and labs. They are not part of the figures at the top.

Hand-in or lab

LAB1 Laboratory Work

  • G
  • U

Main rounds

90% passed · 50 students

  • 89%18
  • 80%15
  • 100%17

Later rounds

100% passed · 5 students

  • 100%5
  • Pass a later round: 100% (5 students), compared with 90% (50 students) on the ordinary exam

Hand-in or lab

PRA1 Project Work

  • G

Main rounds

100% passed · 45 students

  • 100%16
  • 100%13
  • 100%16

Later rounds

100% passed · 5 students

  • 100%5
  • Pass a later round: 100% (5 students), compared with 100% (45 students) on the ordinary exam

About the course

Aim

The course gives knowledge on the algorithms and estimation problems used to infer 3D structure from images. This includes both the mathematics used, and how these are put into practice in algorithm implementation.

After the course, the students should be able to:

  • Goal 1: explain and use algorithms for robust estimation of geometric entities
  • Goal 2: explain and use algorithms for dense correspondence fields and triangulation in stereo images
  • Goal 3: explain and integrate components for structure from motion
  • Goal 4: explain and integrate components for densification and visualisation of 3D models

Content

This course teaches methodology related to the goals listed above, with focus on the following:

  • Robust estimation for geometry (RANSAC)
  • Non-linear least squares estimation, and robust error norms
  • Geometric estimation problems such as absolute and relative camera pose, and minimal cases for solvability.
  • Structure from motion, bundle adjustment
  • Estimation and visualisation of dense correspondence fields
  • Triangulation of correspondences

The contents are introduced in a lecture series, and are then put to use in computer exercises and a programming project.

Examiner
Per-Erik Forssén
Scheduled hours
101 h
Main field of study
Datateknik, Elektroteknik
Syllabus on Studieinfo ↗