Computer Vision for Augmented Reality

Workshop, Reykjavík University, 23th and 26th of February 2018

  • 23.2.2018 - 26.2.2018, 14:00 - 18:00

Augmented reality is an exciting application domain, showing great promise and commercial potential in the field of mobile devices and wearable computing. But how does it work behind the flashy graphics? Within the workshop I will present several concepts of the field that does the hard work behind the curtain – computer vision. We will look at some basic image processing algorithms and how they may be applied to detection of reference markers as well as more advanced approaches using key-points and robust matching. All this will be tied together into AR context with one of the cornerstones of the classical computer vision, the projective geometry. The workshop will consist of lectures that will describe theoretical concepts and practical work where you will be able to test the localization algorithms and combine the camera positioning with computer graphics pipeline in real life using a laptop camera.

Instructor Luka Čehovin Zajc received his PhD at the University of Ljubljana, Faculty of Computer and Information Science in 2015. He works in Visual Cognitive Systems Laboratory at the same faculty as an assistant professor. His research interests are computer vision in relation to intelligent robotics and human-computer interaction. He is involved in various courses on computer vision, mobile robotics and multimedia systems. He is one of the founding members of a highly successful Visual Object Tracking initiative that works on standardization of visual tracking algorithm evaluation.

Dates and group size The workshop will take place in the week between 23rd and 26th of February 2018 from 2PM to 6PM.

Group size is limited to 15 students.

Overview and requirements Day 1 (Friday, 23th of February, 2PM to 6PM):- Introduction
- What is Computer Vision
- Image Processing Basics
- Camera Model and Homography
- Practical work (detection of binary marker)

Day 2 (Monday, 26th of February, 2PM to 6PM):
- Key-points and stable regions
- Robust matching using RANSAC
- Practical work (detection of textured marker)
- Summary

You should bring your laptop and a camera (an integrated laptop camera should be fine, but if you have an external USB camera that will be even better as you will be more flexible). In order to optimize the use of our time it is strongly, strongly encouraged that a C++ compiler / IDE (e.g. Visual Studio) is already installed on the laptop together with the OpenCV library (version 3.0+). It is also recommended that you test your OpenCV installation and IDE configuration by running an some examples. If you are familiar with Cmake build system, you can use it as well. Details about the graphics engine and additional resources regarding installation of OpenCV will be available one week before the workshop.

Application and further questions To apply to the workshop submit a form on 

Application deadline is 8th of February 2018. If you have any additional questions, please contact the instructor: luka.cehovin (AT)