The 70th Heidelberg Image Processing Forum will focus on “Machine Learning and Ground Truth for Image Processing”. New technologies such as deep learning raise new questions: Why are these procedures so successful? What characteristics must data sets have so that learning processes can be successfully developed with them? From a legal point of view, what has to be considered when creating data sets? And how can learning methods be used on modern embedded systems?

70th Heidelberg image processing forum on Machine Learning

Our CTO Dr. Ralf Lay will give a half hour lecture on the topic “Deep Learning with FPGAs – Inline Object Classification in the Production Cycle”. He will discuss the question of how an image processing system that performs its classification via deep learning achieves the necessary data throughput. The FPGA technology with its high parallelism offers a sufficient platform for this. Using practical applications, Dr. Lay explains what needs to be considered, what potential and limitations exist, and what implementations look like with the VisualApplets development environment, which also includes image pre-processing, image post-processing, and signal control.

Presentation (in German) by Dr. Ralf Lay, CTO Silicon Software GmbH, at 4.30 p.m.:
“Deep Learning with FPGAs – Inline Object Classification in the Production Cycle”

With three events per year the Heidelberg Image Processing Forum as an initiative of the HCI (Heidelberg Collaboratory for Image Processing) at University of Heidelberg presents current progress in the digital image processing. Well-known scientists compare notes on how new developed image processing methods can be applied in practice. Co-organizer is the Max Planck Institute for Intelligent Systems.

Dr. Ralf Lay, CTO Silicon Software

70th Heidelberg Image Processing Forum
2 October 2018, 11.00 a.m. to 5.00 p.m.
(followed by a guided tour of the institute until 6.00 p.m.)
Max Planck Institute for Intelligent Systems, Tübingen

http://www.bv-forum.de