PRODUCTS Embedded vision
Smarc 2.1 computer-on-modules Congatec, an edge computing vendor, previewed a low-power computer-on-module for industrial edge analytics, embedded vision and AI at the Embedded World digital show. Te Smarc 2.1 modules are based on the NXP
i.MX 8M Plus processor. Technical highlights of the Arm Cortex-A53-based quad-core processor platform include a neural processing unit and an image signal processor for parallel real-time processing of images and video streams from the two Mipi-CSI camera interfaces. Te Smarc module is available with
application-ready 3.5-inch carrier boards, as well as Basler camera and AI software stack support. Vertical markets for these modules include smart farming and industrial manufacturing, retail, transportation, smart cities and smart buildings. Te neural processing unit adds 2.3 TOPS of AI computational power to the four multi-
Boson for Framos carrier board Framos and Connect Tech have built a board-level solution that connects the Framos sensor module ecosystem to a carrier board for production deployment of Nvidia Jetson solutions. It is designed for Jetson Nano and Jetson Xavier NX edge AI applications. Using Boson for Framos, customers ready
to migrate applications from Jetson developer kits to production benefit from a size reduction, plus double the interfaces – three 4-lane or four 2-lane Mipi CSI-2 connectors. Direct camera integration to the carrier
board eliminates the need for an adapter board and cuts the overall hardware stack height. Boson for Framos comes standard with software hooks in the board support package, that connects select cameras in the Framos sensor module ecosystem to Nvidia’s JetPack SDK to minimise software development.
Proc10N and Proc1C10N FPGA modules New from Gidel are the Proc10N and Proc1C10N, an FPGA module and PCIe board based on the Intel Stratix 10 NX. Tese compact solutions give a fresh option for complex AI and other compute-intensive tasks requiring high data rate and low latency. Te Stratix 10 NX incorporates new Tensor
blocks for fast AI and vector processing. Te Tensor blocks integrated with HBM2 memory offer excellent computation performance and memory bandwidth. Te Stratix 10 NX’s capabilities make it
suitable for high-bandwidth real-time uses, such as AI, video analytics, IT-security, 5G and radar. Its compact design make it well suited for small, low-power embedded systems. Te Proc10N module provides 1,600 Gb/s I/O
throughput, Tensor compute blocks capable of 143 INT8 TOPS, and 400 GB/s access to HBM2
www.imveurope.com | @imveurope
8GB DRAM for fast AI, convolutions, FFT, encoders, filtering, linear algebra operation and beyond. Te Proc10N module measures 97.4 x 101mm and offers full I/O interface customisation via a user or Gidel carrier board. Customised solutions can be developed
using the Proc10N module, a carrier board, and Gidel’s ProcVision development suite. ProcVision simplifies development on FPGAs and improves system integration and reliability.
https://gidel.com/
purpose Arm Cortex-A53 processor cores. Te ISP processes full HD video streams with up to three-times 60fps for video enhancement. Te DSP enables local speech recognition without any cloud connection. Te Cortex-M7, which can also be used as a fail-safe unit, provides real-time control together with a time synchronised networking capable Ethernet port. Te modules have 6GB LPDDR4 memory,
and can drive up to three independent displays.
www.congatec.com
VC PicoSmart Measuring 22 x 23.5mm, Vision Components says its new VC PicoSmart board is one of the world’s smallest embedded vision systems. Te board camera contains all the components needed for image processing, including a 1-megapixel CMOS sensor with global shutter, FPGA module, high-end FPU processor and memory. An interface board can be connected via the FPC connector. VC PicoSmart is the perfect base for rapid
development of cost-effective, application- specific and compact vision sensors. Vision Components’ VCRT operating system enables real-time image processing, for instance for object recognition, position control, barcode reading, as well as web edge and level control. Vision Components presented the board
camera at the Embedded World trade show, along with new Mipi camera modules, including a module with the latest Sony swir sensors.
www.vision-components.com
Connect Tech is an elite partner in the
Nvidia partner network and has been developing hardware solutions for Nvidia Jetson since its inception. Framos joined the partner network in 2020 as a preferred partner and offers Jetson customers a broad sensor module selection range for many applications.
www.framos.com
Embedded processing kit At the Embedded World show, Basler presented its embedded vision processing kit, which includes various interfaces for connecting different camera types. Te in-house developed board benefits from a flexible system-on-module and carrier board approach based on NXP’s
i.MX 8M Plus system-on-chip (SoC). Tis SoC has a neural network processor unit and an image signal processor for machine learning. Te kit includes the Pylon camera software
suite, which provides drivers for all types of camera interface, simple programming interfaces, and a set of tools for camera setup. For vision applications, Bcon for Mipi, GigE Vision and USB3 Vision are available as interfaces. Standard interfaces include: HDMI, GPIOs, SPIs/I2Cs, LVDS, UART, CAN, USB2.0, wifi, and Bluetooth. Customers can configure cameras with a wide
range of sensors and performance specifications from Basler’s portfolio. Te design, geared towards industrial use, can be adapted for any vision project with little effort. General availability is planned for July.
www.baslerweb.com
APRIL/MAY 2021 IMAGING AND MACHINE VISION EUROPE 31
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36
