Seminars & Projects

Project and Seminar Courses

The Bachelor and Master Seminars introduce students to working critically with the literature in pattern recognition and related areas in preparation for their Bachelor/Master theses. Students are expected to write a term paper and make a presentation of their chosen topic. 
  • 89-1311 Bachelor Seminar: Grundlagen der Mustererkennung
  • 89-1311 Bachelor Project: Grundlagen der Mustererkennung
  • 89-1371 Master Seminar: Topics in Pattern Recognition
  • 89-1381 Master Project: Development of Pattern Recognition and Image Understanding Systems
You receive credit for your project results and seminar presentations.

Areas and Topics

There are a number of different areas in which we supervise seminars and projects.

computer vision and pattern recognition
  • fast parallel computer vision and pattern algorithms in OpenCL, CUDA, OpenMP, Map-Reduce
  • fast, large-scale approximate string searching for text mining and OCR
  • large scale benchmarking
  • 3D computer vision and stereo matching
  • x-ray image processing and object recognition for airport security
document analysis and forensics
  • learning of non-linear image processing filters for document images
  • 2D image segmentation and analysis
  • undetectable steganography
usability and user interfaces
  • automatic evaluation of user interfaces
  • instrumentation and statistical analysis of live user data
  • evaluation of video recordings of user data using image analysis
  • automatic visual understanding of user interfaces
simulation and virtual worlds

This is a new application area for techniques developed in our group.  We'll be offering courses in the coming years.
  • 3D capture of real-world objects for building models for virtual worlds
  • large scale efficient simulation algorithms for virtual worlds
  • large scale efficient simulation of 3D structures
  • distributed computation for simulation, virtual worlds
  • artificial intelligence in virtual worlds (e.g. SecondLife, OpenSim)

Open Source Projects

Students have the opportunity to get involved in one of two open source projects with many real world applications and interested user communities.  These projects also provide potential opportunities for internships and funding for graduate studies:
  • The OCRopus system (ocropus.org) is a new open source OCR system for capturing books and other documents. The goal is to make books and literature widely available and accessible, in particular out-of-copyright classical literature and historical documents. OCRopus is sponsored by Google and the TextGrid project, and we collaborate with the Internet Archive and other academic libraries and institutions.
  • The DECAPOD system (decapod-project.org) is a system for camera-base document capture, intended to allow small libraries and museums to capture and preserve cultural treasures. The DECAPOD project is sponsored by the Mellon Foundation and works in collaboration with the University of Toronto. 
Comments