Deep Learning for Visual Computing

Course Details


Christopher Pramerdorfer
Michael Reiter
Martin Kampel




Vorlesung mit Übung (VU)



This course will not be available in WS17.

  • When: Thu 13:15 – 14:45
  • Where: schedule

Course description

This lecture covers Deep Learning for automatic image and video analysis, such as classifying images into categories or detecting and distinguishing persons. Deep Learning has recently lead to breakthroughs in these fields; in certain problems, the performance of current methods based on this technology is similar or even better than that of humans – a novelty in this field.

Deep Learning in action (image from

The goal of this lecture is to provide a comprehensive introduction to this exciting branch of machine learning. We will focus on Convolutional Neural Networks, the most popular model for image analysis, but also cover related models such as Stacked Autoencoders.

You will apply what you’ve learned in the exercise part of this course, which consists of several assignments that must be handed in by each student (groups of two are fine). You can work on these assignments on your own computer if you have a decent GPU with CUDA support. We will provide remote access to a dedicated GPU server for those who don’t.


This is a course for Master’s students, so students are expected to have basic knowledge of mathematics and statistics, image processing, and machine learning. We will briefly recap some basics as part of the first few lectures.

For the exercise part, students are expected to be able to program in Matlab, Python, Java, or Lua. Any of these languages is fine, although we favor Python.


There will be a written exam that covers the lecture part (50% of the grade). The exercise part is also worth 50% of the grade.

Lecture Slides (WS16)



If you have completed the assignments and want to take the exam, please contact Christopher. A list of questions as well as exam information is available here.