At VISION 2018, booth 1C72, we will focus on CNN-based deep learning hardware and software on FPGAs for real-time applications with high bandwidths in object and feature classification. The close interaction of our hardware and software is suitable for the integration of neural networks of different depth and size for industrial requirements such as inline inspection, robotics and autonomous driving as well as for non-industrial sectors such as medical technology, agriculture and research. In addition to our frame grabber hardware, embedded devices such as cameras and vision sensors are also compatible with our software, the graphical development environment VisualApplets, and can be equipped with individual image processing applications.
A demonstrator will show a high-speed deep learning application on FPGA, detecting six different defect classes on reflecting metallic surfaces with very high classification accuracy.

CNN frame grabber

New Camera Link high performance frame grabber, also suited for Deep Learning

New Frame Grabbers

For deep learning as well as other high-speed applications in image processing, we present new Camera Link and CoaXPress frame grabbers:

  • microEnable 5 marathon VCLx with a data input bandwidth of up to 850 MB/s, for complex applications, programmable, and available as microEnable 5 marathon deepVCL with special license for CNN projects: More Information
  • microEnable 5 marathon ACX/VCX-QP with resp. four ports, ACX-DP with two ports and ACX-SP with one port (from CXP1 up to CXP6): More Information
  • LightBridge 2 ACL/VCL as a compact external frame grabber, deployable with electrical or optical ThunderboltTM 2 cables: More Information

At our booth we will inform about planned frame grabbers with next generation interfaces like CXP-12, 10 GigE and NBASE-T.

New Software Version

With VisualApplets Version 3.1 the license scheme for VisualApplets Expert has been adapted. In addition to purchasing the entire solution, the Parameters and Debugging modules are now also available separately. Also new in VisualApplets Expert is a console with built-in Tcl interpreter. This allows users to automate the design process. For example, complex processes such as the simulation of thousands of images with different parameter settings can be performed automatically via script.
With the configurable “Update Checker”, users can automatically search for the latest VisualApplets version at each program start or manually via the VisualApplets GUI. Preconfigured VisualApplets designs (e.g. Normalized Cross Correlation, Exposure Fusion and Distortion Correction) complete version 3.1.
More Information

New version VisualApplets 3

Many new features in VisualApplets and the extension VisualApplets Expert

Meet us on the following events!

Press Conference with Subsequent Tour of the Fair

Tuesday, November 6, 11:00 a.m – 12:30 p.m.
Our managing director Dr. Klaus-Henning Noffz presents the market figures of the VDMA (15-20 minutes).

Forum Industrial VISION Days – Panel Discussion (in English)

“The Future of Vision: Are we entering a new era?”
Tuesday, November 6, at 2:00 p.m.
“Galerie L-Bank Forum” in Hall 1, Booth Z75
Moderated by Joachim Hachmeister, editorial dept. inspect
Our Managing Director Dr. Klaus-Henning Noffz will participate as Chairman of the VDMA Machine Vision group.

More Information

Forum Industrial VISION Days – Speech Deep Learning (in English)

Thursday, November 8, 3:00 – 3:30 p.m.
“Galerie L-Bank Forum” in Hall 1, Booth Z75
Our head of development Markus Grebing gives a lecture about “VisualApplets 4 Deep Learning – CNNs enter Factory Floor”.

More Information

We look forward to speaking with you on VISION 2018!
Appointment and Tickets Request

Silicon Software on VISION 2018
Hall 1, Booth C72
November 6-8, 2018 in Stuttgart
https://www.messe-stuttgart.de/vision/en/