Universität Bonn: Autonomous Intelligent Systems Group  Comouter Science Institute VI: Autonomous Intelligent Systems

Lab Vision Systems - Learning Vision Systems on Graphics Cards  MA-INF 4308


Prof. Dr. Sven Behnke, Simon Bultmann

weekly, dates by agreement, work on the final project will be possible in the lecture-free time in summer

Final project due: September 30.

  - If you plan to participate, please contact bultmann@ais.uni-bonn.de
  - Places are limited and will be assigned randomly among attendees of the introductory meeting.

Content

Due to the availability of general purpose programming interfaces like CUDA, the immense speed of graphics cards can be put to work for a multitude of parallel tasks. Algorithms for the analysis of images mostly work independently on different regions of an image. These algorithms are therefore inherently parallel and can greatly profit from parallel hardware.

Speedup factors in the order of two magnitudes make it possible to process and extract information from huge datasets, for example the images of the ImageNet Large Scale Visual Recognition Challenge. When experimenting with learning algorithms, the experiment duration is drastically reduced.

In the Lab, we learn how to implement learning algorithms from the area of visual pattern recognition and accelerate them using graphics processing unit (GPU). We will implement learning algorithms with the help of PyTorch which is a popular deep learning framework. The lab will have weekly meetings. We'll agree on a time slot for these meetings in the first session.


Prerequisites

  • Programming skills in python
  • Knowledge in the area of artificial intelligence and machine learning would be helpful.
  • Recommended: have completed at least one of the following lectures
    • MA-INF 4111 Intelligent Learning and Analysis Systems: Machine Learning 
    • MA-INF 4204 Technical Neural Nets 
    • MA-INF 2313 Deep Learning for Visual Recognition

Course Material

Slides and Assignments


Application Domain: Pascal Object Recognition Challenge

Pascal Object Categorization Challenge

Our three Titan CUDA computers with a total of 34.560 CUDA cores and 96GB DDR5 RAM

CUDA-Supercomputer

University of Bonn, Institute for Computer Science, Computer Science VI - Intelligent Systems and Robotics    Impress    Data Privacy Statement