Computer Vision Systems Programming (UE)

Course Details


Martin Kampel




Übung (UE)



  • When: Wed 10.00 – 12.00 c.t.
  • Where: Seminarraum 183/2 (map)

Course description

This course encourages students to select and implement a computer vision project of their choice. Students are free to develop in any programming language they like and to use any publicly available library they want. The only requirement is that the effort for developing the chosen application is in line with the ECTS of this course. The goal is to encourage students to investigate a selected computer vision topic in detail, and to allow them to improve their computer vision programming skills.

Sensor hardware is provided. The available hardware includes Kinect sensors (both versions), a thermal imaging camera, a network of multiple IP cameras with overlapping views, and portable Android devices with cameras.

Participants are encouraged to select a computer vision problem according to their interests. Some examples are presented as part of the first lecture, and the lecturers are happy to help participants choose their topics. It is also possible to implement an internship topic, partially or fully in groups.

An example for a project that was developed for this course is Wurstify.


This is not a general programming course; students are expected to be able to develop software on their own, and they should be familiar with a programming language suitable for computer vision development (e.g. Matlab, Python, C++). Basic image processing and computer vision knowledge is expected. Experience in computer vision development is recommended but not required.


Project 01a: Shot Type Classification – dl-based (Koch Thomas)

Project 01b: Shot Type Classification – face-based

Project 02a: Camera Movement Classification: A fundamental base for automatic video analysis –> optical flow (Pescoller Katharina)

Project 02b: Camera Movement Classification: A fundamental base for automatic video analysis  –> DL-based  (Patrick Link)

Project 04: Frame Border Detection in historical videos (Pointner Bernhard)

Project 05: Evaluation of Speech-to-text APIs

Project 06: Evaluation of Object Detection algorithms in historical videos (Marvin Burges)

Project 07: Object-Camera distance estimation in videos (Jafari-Sahamieh Hamed)

Project 08: CV algos: mobile versus server based (Dominik Scholz)

Project 09: Mobile face recognition / Biometric Match (Pointner Michael)

Project 10: Mobile: combining face recognition & finger print (David Ammer)

Project 11: Avatar”steuerung” mittels Behavior detection (Felix Ginzinger)

Project 12: Edge Analyse mit Hanwha Wisenet sowie Axis (mit ARTPEC 6/7 Chipsatz) Kameras

Project 13: Deep Learning Klassifikationsservice auf Basis von
OpenVINO (Jakob Troidl)

The topics are further described in cvp2019_topics_v5.


The initial, midterm, and final presentations account for 5%, 5% and 10% of the grade, respectively. The project, which must include a written report, is worth 80% of the grade.

Associated lecture

We recommend the associated lecture that covers software and resources for computer vision development as well as selected computer vision applications.


see in  TISS: 183.586 Computer Vision Systems Programming  –> Einzeltermine