EMBEDDED MACHINE VISION - STEMMER IMAGING · Texas Instruments Keystone •ARM + DSP Altera (Intel) Cyclone V •ARM + FPGA Start with evaluation board Make your own board ARM cores

A STEMMER IMAGING PRODUCT

EMBEDDED MACHINE VISION

RUPERT STELZ

A STEMMER IMAGING PRODUCT

1.Classification of embedded machine vision systems

2.Details on ARM based embedded systems

3.Solution

CONTENT

A STEMMER IMAGING PRODUCT

CLASSIFICATION OFEMBEDDED MACHINE VISION SYSTEMS

Smart Camera Embedded PC System On Chip Deep Embedded

Camera, processing and I/O in one housing

Point and click environment

No programming required

Connection to factory automation

Limited in sensor support

Limited in functionality

A STEMMER IMAGING PRODUCT

CLASSIFICATION OFEMBEDDED MACHINE VISION SYSTEMS

Smart Camera Embedded PC System On Chip Deep Embedded

External camera(s), all sensors

Embedded Windows operating system

Windows IoT or Windows Embedded

Free programmable using machine vision

libraries

Connection to factory automation by adapter

cards or specific network cards (stacks)

‘Traditional’ Windows platform with all pros

and cons

A STEMMER IMAGING PRODUCT

CLASSIFICATION OFEMBEDDED MACHINE VISION SYSTEMS

Smart Camera Embedded PC System-on-a-Module Deep Embedded

External camera(s), all sensors

Linux operating system

Free programmable using machine vision

libraries

Low power, simple duplication

Connection to factory automation limited

C/C++ programming

A STEMMER IMAGING PRODUCT

CLASSIFICATION OF EMBEDDED MACHINE VISION SYSTEMS

Smart Camera Embedded PC System On Chip Deep Embedded

Fully integrated machine vision system

Maybe no operating system at all

Not programmable, just one application

Low power, mass market

Connections as defined by the system

Mass market product

A STEMMER IMAGING PRODUCT

WHY ?

Size ?!

Price ?!

Maintenance ?!

Scalability ?!

A STEMMER IMAGING PRODUCT

DETAILS ON MODULAR EMBEDDED

Of the shelf hardware Special hardware Operating system Deep Embedded

ARMV7-A as the 32bit platform

ARMV8-A as the 64bit platform (A57)

Recommended

• Gbit LAN, USB3

Often USB2 and 100Mbit LAN only

(Raspberry Pi)

Gbit LAN often connected via USB

resulting in low network performance

Odroid XU4/C2/N2

NVIDIA Nano / TX2 / Xavier

NXP i.MX x Series (i.MX 8)

Raspberry Pi

X86

ESPRESSObin

A STEMMER IMAGING PRODUCT

DETAILS ON MODULAR EMBEDDED

Of the shelf hardware Special hardware Operating system Available software libraries

Texas Instruments Keystone

• ARM + DSP

Altera (Intel) Cyclone V

• ARM + FPGA

Start with evaluation board

Make your own board

ARM cores available in hard- and software

floating point

A STEMMER IMAGING PRODUCT

DETAILS ON MODULAR EMBEDDED

Of the shelf hardware Special hardware Operating system Available software libraries

Ubuntu LTS (legacy)

CVB ships with debian (deb) package

• .sh installer for Ubuntu

• Resolves external dependencies

Yocto

• Minimum footprint Linux

• 32bit and 64bit ARM

• CVB ships with Tarball

Full documentation about installing CVB

on Linux and the installation packages are

available here:

https://www.commonvisionblox.com/en/do

wnload-cvb-2017-linux/

A STEMMER IMAGING PRODUCT

DETAILS ON MODULAR EMBEDDED

Of the shelf hardware Special hardware Operating system Software

Jetson shipped with:

• Cuda libraries and samples

• OpenCV libraries and samples

Some have OpenCL Support

Image Manager for Linux

• Supports GigE-Vision cameras

• Supports USB3-Vision cameras

• Image access and display

• Python support and C++ interfaces

Machine learning

• OCR with up to 40.000 characters per

• Precise recognition of object orientation

and size

GigE-Vision Server

• Build your own GigE-Vision device

• Add user defined features to your virtual

camera

CVB as be building base

• In combination with OpenCV

• In combination with NumPy/SkiPi

• In combination with TensorFlow

A STEMMER IMAGING PRODUCT

CLASSIC DEVELOPMENT

Compiler / Tool chain CMake Qt Python / C++

GNU Compiler

Cross platform development with CMake

Qt (QML) for GUI Development

CM

ake

Vis

ual

Stu

dio

GN

U

Makefile

Execut-

able

Execut-

able

A STEMMER IMAGING PRODUCT

SCRIPTING FOR RAPID PROTOTYPING

Compiler / Tool chain Python / .NET C++ Extensible

Use scripting to speedup development

Same script runs on Windows and Linux

Building the bridge to other infrastructures

Intuitive cross platform API

Easy migration to native applications

A STEMMER IMAGING PRODUCT

DEVELOPMENT IN C++

Compiler / Tool chain Python / .NET C++ Extensible

C++ class library using state of the art technology

Using the same source code

Very close mapping to Python and .Net API

Develop using scripts, deploy native

Header only lib using native components

Debug under Windows, compile for ARM

A STEMMER IMAGING PRODUCT

DEVELOPMENT INFRASTRUCTURE

Compiler / Tool chain Python / .NET C++ Extensible

Convenient scripting in Python

Class structure is the same in C++

Up-to-date programming interface

A STEMMER IMAGING PRODUCT

INTERFACING TO CAMERAS

GigE Vision

USB3 Vision

MIPI CSI

SLVS-EC

…

GenICam GenTL

V4L

GStreamer

Proprietary

…

CVB acquisition interface

A STEMMER IMAGING PRODUCT

FACTORY INTEGRATION

CVB will support Open Process Communication

- Unified Architecture

OPC-UA

OPC-UA connects devices in the factory floor

VDMA is hosting a standard group called OPC-

UA Vision

STEMMER IMAGING is a contributing member

A STEMMER IMAGING PRODUCT

MORE INFORMATION

COMMONVISIONBLOX.COM

FORUM.COMMONVISIONBLOX.COM

www.commonvisionblox.com/en/download-cvb-2017-linux/

A STEMMER IMAGING PRODUCT

THANK YOUFOR YOUR ATTENTION

Your contact

Rupert Stelz

STEMMER IMAGING AG

r.stelz@stemmer-imaging.de

www.stemmer-imaging.de

© Copyright STEMMER IMAGING AG. All rights reserved. All texts, images, graphics, sound-, video- and animation files, as well as

their arrangements are copyright protected. Reprint, processing and duplication for commercial purposes or use on websites are forbidden.

Some STEMMER IMAGING pages contain images that are subject to copyright of the respective owner.