FPGA technologies have made a real evolutionary leap, requiring new approaches and solutions from both hardware and software developers. At the FPGA Congress 2019 from 21 to 23 May such approaches will be presented, discussed and evaluated in more than 100 lectures and hands-on tutorials by more than 70 high-quality international speakers in a total of ten main topics. Participants also benefit from test developments on provided boards and computers. An accompanying exhibition will show applications where we have practically implemented a high-speed deep learning application for a difficult inspection environment.

Also at this event the topic Deep Learning on FPGAs moves more and more into the center of attention, as FPGA technology offers the best solution for the inference of neural networks in the industrial environment. FPGA-based inferences for complex applications combine high computing power with low power requirements, guaranteed latency and long-term availability. The technology is also suitable for embedded image processing systems and especially for inline inspection with its real-time requirements. Similar applications with conventional GPUs can only be realized with a compromise in speed, latency, detection rate, energy efficiency or overall system costs. The latter can be reduced by smaller networks and FPGAs without sacrificing recognition accuracy, as well as by saving system components such as GPUs and the use of smaller CPUs.

With FPGAs Real-time Results for High-resolution Images

The FPGA approach provides real-time results even for high-resolution images: from image acquisition to image pre-processing (data reduction, image optimization) directly to result images. Numerous applications, including embedded ones, can be implemented, including factory automation, robotics, medical technology, safety, transport (vehicles and drones), systems for driver assistance and autonomous driving, and electronic components.

Deep Learning – Suited for Many Industries

Lots of Topics Concerning FPGAs

Our Head of Purchasing & Production, Dr. Holger Singpiel, will talk about how image processing tasks with difficult environmental conditions can be implemented safely, quickly and cost-effectively with Deep Learning. When it comes to object and feature classification, deep learning is superior to classical methods for difficult inspection tasks with many feature or environment variables or patterns that are difficult to describe analytically.
Speech: „Deep learning high performance solution on FPGAs“ (in German language)
23rd May 2019, 1:30 – 2:15 p.m

  • Develop powerful deep learning based image processing solutions on FPGAs quickly and efficiently with the graphical programming environment VisualApplets
  • Easy porting of deep learning applications to other (embedded) FPGA hardware such as cameras and vision sensors
  • Reduced development times and total system costs through FPGAs.

For our multiple awarded FPGA programming environment VisualApplets we offer certifications to interested companies worldwide to become our sales or development partner. Currently, certified VisualApplets Competence and Design Centers exist in 34 countries worldwide.
If interested, get in touch with Mr. Singpiel directly during the event or with our Sales under Tel: +49 621 789507-39 or by E-mail: sales@silicon-software.de.

FPGA Congress 2019
21st-23rd May 2019
NH Hotel München Dornach
Einsteinring 20, 85609 Dornach-München
http://www.fpga-kongress.de