Skip to content

Qualcomm Edge AI Vision kit combines Qualcomm QCS610 SoC & Sony IMX415 4K ultra low light camera

Qualcomm Edge AI Vision kit combines Qualcomm QCS610 SoC & Sony IMX415 4K ultra low light camera

e-con Systems recently launched the qSmartAI80_CUQ610, a Qualcomm Edge AI vision array based on the Qualcomm QCS610 octa-core Cortex-A76/A55 processor and incorporating a camera module based on the Sony STARVIS IMX415 4K Ultra Low Light Sensor.

The kit consists of a QCS610 module with 4 GB RAM and 16 GB eMMC flash memory and a bus board designed by VVDN Technologies, as well as an e-con Systems 4K camera module, designed to power machine learning and deep learning models at the edge.

Qualcomm Edge AI Vision suite

qSmartAI80_CUQ610 “Qualcomm Edge AI Vision Kit”:

  • system on unit
    • SoC – Qualcomm QCS610 Octa-core processor with 2x Kryo 460 Gold cores @ 2.2 GHz (Cortex-A76 class), 6 x Low Power Kryo 430 Silver @ 1.8GHz (Cortex-A55 class), Adreno 612 GPU @ 845 MHz, with OpenGL ES 3.2, Vulkan 1.1, OpenCL 2.0, Hexagon DSP with Hexagon Vector eXtensions (HVX), Spectra 250L ISP, 4Kp30 VPU H.265; Note: Qualcomm QCS410 SoC is available upon request
    • System Memory – 4 GB LPDDR4X SDRAM, 1866MHz
    • Storage – 16GB eMMC Flash
    • Dimensions – 55 x 35.5 mm
  • 4K camera module
    • Sensor – Sony STARVIS IMX415, up to 4K UHD (3840 x 2160)
    • Output format – 10-bit RAW Bayer up to 4Kp30
    • Focus Type – Fixed Focus
    • Optical coordination – 1/2.8″
    • Shutter type – electronic rotating shutter
    • Pixel size – 1.45 µm x 1.45 µm
    • Mount – M12 (S-Mount)
    • DFOV – 115.90° with eyepiece provided by e-con . Systems
    • Dimensions – 30 x 30 mm
  • conveyor plate
    • Storage – MicroSD card socket
    • Display I/F – HDMI, 4-PAY MIPI DSI Connector
    • Camera I/F – 3 x 4-lane MIPI CSI Connectors
    • My voice
      • Low power stereo codec with headphone output
      • 2x 2-bit MI2S port, 1x SLIMbus interface for BT audio, 1x Single-track SoundWire interface
    • Networks
      • Gigabit Ethernet
      • Dual Band WiFi 5 1 in 1 and Bluetooth with 1x u.FL antenna connector
    • USB – 1 x USB 3.1, 1 USB 2.0 Type-C, 1 x USB 2.0 . host port
    • Power Supply – 12V DC
  • Compatibility – RoHS

VVDN QCS610 SUM bus board

The kit also comes with a coaxial cable to connect the camera module to the bus board, the camera lens, a USB Type A to Type C cable for connecting the kit to a host/development device, and a 12/3A power adapter.

On the software side, e-con Systems provides Yocto-build Linux image with GStreamer-based camera application for “HDMI stream”, H.264 network streaming over Ethernet and WiFi, snapshot image capture in JPEG and RAW Bayer, H.264 video recording And adjust camera features like white balance, gain, exposure, etc. Engineers can take full advantage of the AI ​​capabilities of the suite using the Qualcomm Neural Processing SDK for AI, Neural Network API (TFLite Models), GStreamer Plugin (DLC Models), and GStreamer TFLite Plugin. Software and hardware documentation is available after free email registration.

QCS610 camera power supply board

e-con Systems expects its customers to use the suite to develop service robots, telemedicine robots, telepresence robots, smart departures and carts, smart signage and kiosks, meter and body cameras, and automated sports broadcast systems

We’ve written about System on Modules for years, but I can’t remember covering products from VVDN Technologies, and based on the information on their LinkedIn profile, they’re not a small company with over 3,000 employees. Going to the company’s website explains why: They don’t have products, and instead, they provide engineering and manufacturing services to other companies. A quick search on CNX software reveals that I quickly mentioned the NXP i.MX 6UL device built by VVDN in 2016 for Android Things.

Back to the main topic. e-con Systems qSmartAI80_CUQ610 QCS610 is now available, but you will need to contact the company for pricing. More details, including software and hardware documentation, can be found on the product page.

Source link