All News | Boards | Chips | Devices | Software | LinuxDevices.com Archive | About | Contact | Subscribe
Follow LinuxGizmos:
Twitter Facebook Pinterest RSS feed
*   get email updates   *

Microsoft’s Azure-focused, 8MP smart AI camera runs Linux on Qualcomm SoC

Sep 5, 2019 — by Eric Brown — 809 views

Microsoft announced a $249 “Vision AI Developer Kit” with an 8MP, 4K camera that runs Linux on Qualcomm’s 10nm, AI-enabled QCS603 SoC. The kit is aimed at AI edge developers using Azure IoT Edge and Azure Machine Learning.

Some of us old timers are still getting used to Microsoft choosing Linux to power its Azure Sphere platform, as seen in Avnet’s recent Azure Sphere MT3620 Starter Kit. Now Microsoft has found some new partners to help it launch a Linux-based AI edge smart camera. The Vision AI Developer Kit combines Qualcomm’s Vision Intelligence Platform, which is baked into its QCS603 SoC, with Microsoft’s Azure IoT Edge service, built on the Azure IoT Hub. The platform also includes Vision Studio and Azure Machine Learning (AML) for building, training, and deploying machine learning models.



Vision AI Developer Kit

The Vision AI Developer Kit was originally announced in May 2018 as a collaboration with Qualcomm, which had recently announced its Qualcomm Vision Intelligence Platform. This hardware/software platform for deep learning was deployed its 10nm fabricated octa-core QCS605 and quad-core QCS603 SoCs.

The Vision AI Developer Kit is marketed and distributed by eInfochips and sold by Arrow, which has used eInfochips before for its self-branded AI-ML Board and Thor96 SBCs. The kit is built by Altek, which also makes a Linux-driven, QCS603-based IPC603 camera reference design based on the QCS603.

The Vision AI Developer Kit is billed as an “end-to-end Azure enabled solution with real-time image processing locally on the edge device, and model training and management on Azure.” Microsoft claims that the kit lets you “deploy vision models at the intelligent edge in minutes, regardless of your current machine learning skill level.”


Vision AI
Developer Kit

Three development paths are available: “no code” using Custom Vision; Azure Cognitive Service, custom models with Azure Machine Learning; and the Visual Studio Code IDE. Aimed at novices, Custom Vision walks users through the process of uploading data, training, and deploying customer vision models including image tagging. With the Vision AI Developer Kit, users can then use Azure IoT Hub to deploy a custom vision model directly to the kit.

More advanced users can try the Azure Cognitive Service with visual drag and drop tools for AML development . Reference implementations provided in Jupyter notebooks “walk data scientists through the steps to upload training data to Azure Blob Storage, run a transfer learning experiment, convert the trained model to be compatible with the developer kit platform, and deploy via Azure IoT Edge,” says Microsoft.

Finally, more advanced developers can use the kit’s extension for Visual Studio Code, which offers sample Python modules, pre-built Azure IoT deployment configurations, and Dockerfiles for container creation and deployment. Visual Studio Code can also add business logic to existing Azure solutions based on camera input sent via Azure IoT Hub to transform the image data into normalized data streams using Azure Stream Analytics.

 
Qualcomm Vision Intelligence Platform and other hardware details

We covered the higher-end QCS605 version of the Qualcomm Vision Intelligence Platform last November with a story on Intrinsyc’s QCS605-based Open-Q 605 SBC. The quad-core QCS605 has 8x Kryo 300 CPU cores, two of which are 2.5GHz “gold” cores equivalent to Cortex-A75 and the other six 1.7GHz “silver” cores like the Cortex-A55. The quad-core QCS603 on the Microsoft kit is identical except that it offers only 2x of the 1.7GHz “Silver” cores instead of six.

Both Vision Intelligence Platform SoCs feature an Adreno 615 GPU, a Hexagon 685 DSP with Hexagon vector extensions (“HVX”), and a Spectra 270 ISP. All these chips are part of the Qualcomm AI Engine built around the Qualcomm Snapdragon Neural Processing Engine (NPE) software framework, which enables analysis, optimization, and debugging tools for developing with Tensorflow, Caffe, and Caffe2 frameworks.



Vision AI Developer Kit detail views
(click image to enlarge)

The Vision AI Developer Kit is built around Qualcomm’s Altek-built Vision Intelligence 300 Platform, also referred to as the EIC MS Vision 500 smart camera. The device backs up the Yocto Linux driven QCS603 SoC with 4GB LPDDR4. For storage, there’s 16GB eMMC and a microSD slot.

The centerpiece is an 8-megapixel camera with 4K UHD support and a four-microphone array. The device also provides a dual-band 802.11b/g/n radio, audio I/O jacks, an HDMI port, and LEDs. For power, there’s a USB Type-C port and a 1550mAh battery. The device can swivel on its base, enabling multiple mounting options.


AWS DeepLens

The VentureBeat story that alerted us to the Vision AI Developer Kit compares the kit to Amazon’s $249, 4-megapixel AWS DeepLens camera. The DeepLens runs Ubuntu on an Intel Cherry Trail SoC.

 
Further information

The eInfochips/Altek Vision AI Developer Kit is available for $249 on this Arrow shopping page. More information may be found in Microsoft’s Vision AI Developer Kit shipment announcement, as well as the product page and GitHub page. Altek also has its own product page for the kit, which it calls the AI Camera by Altek.

 

(advertise here)


Print Friendly, PDF & Email
PLEASE COMMENT BELOW

Please comment here...