Centric connect.engage.succeed

Azure IoT and the Intelligence Cloud and why we need Intelligence Edge!

Geschreven door Randolph Widjaja - 29 mei 2018

Randolph Widjaja
Last year, I created a PoC environment with Cortana Intelligence Suite, where I used Azure Cognitive Services with Azure Bing Search services, just to check out the capabilities of the Azure Intelligence Cloud. I downloaded the Azure IoT face recognition doorbell software for Raspberry Pi systems from GitHub and connected this to the Azure Cognitive Services for facial recognition. The software unlocks the front door automatically when faces are recognized through the camera. If someone is at your door and is not recognized the door stays locked. Don’t you want to know who has come to your home and if you can trust that person? I know I do.

Azure Cognitive Services

So, with the help of Azure LogicApps I connected the Azure Cognitive Services Bing Search API to my Azure IoT setup. When someone rings my doorbell a picture is uploaded to Azure Blob Storage and processed by Bing Services. The search results are sent to my email address, where I can allow or deny access. I won’t go into the details in this blog, but if you would like to know how I setup and configured my Azure IoT environment, I’ll publish it soon on my own blog site http://www.randolphwidjaja.com.

The results from the Bing Search were nice, but with Bing my face looked the same as that of other people found on the internet (you can see the results below). So, for better reading I converted the search results from JSON to Bing GUI, which came up with the same results however.

This year Microsoft updated its Azure Cognitive Services like Face and Bing Search. This presented me with a new opportunity to test the functionality of Azure Cognitive Services. I assembled my IoT setup again and checked to see whether it would now find my own pictures on the internet. The results are shown below, and I finally found my own face in the results.

Intelligence Edge

Intelligence Cloud is still developing and getting stronger all the time. The only downside of my IoT setup is that the Cognitive Services are located in the datacenters of Microsoft Azure. This means that my IoT setup needs an internet connection to process the data, which can be interrupted and keep my facial recognition doorbell from working. To overcome this problem Microsoft is bringing the AI to the edge where the sensors are located ¬ – to the ‘Intelligence Edge’. This gives the IoT ecosystem more distributed power, more decentralized public cloud features and more offline capabilities.

To move the processing to the edge a couple of things need to be taken into account. The edge needs to have the capacity to process and analyze the data. The sensor devices should be connected to a processing unit capable of analyzing the cognitive features. This can be done with a neural processor, which nowadays can be found in devices like mobile phones, cameras, etc. More and more devices are being made ready for the Intelligence Edge, and the number of devices with neural processing power will be growing in the coming years.

What’s more, Microsoft is entering into partnerships with various vendors to create a bigger community for the Intelligence Edge. One of them is Qualcomm, the company Microsoft partnered with to create the vision AI developer kit. The kit includes a camera and hardware that allows cognitive services and Azure Machine Learning to run locally, without an internet connection, and interpret the video output. I can’t wait to get my hands on one of those devices to try it out.

Sidenote: because of the GDPR privacy laws, Microsoft updated the use of the Search API and display requirements policies. More info can be found on https://docs.microsoft.com/en-us/azure/cognitive-services/bing-entities-search/use-display-requirements.

Sources:

Azure Cognitive Services update
https://azure.microsoft.com/en-us/blog/microsoft-empowers-developers-with-new-and-updated-cognitive-services/

Announcements about the Intelligent Edge at Build 2018
https://www.microsoft.com/en-us/internet-of-things/intelligentedge

GPDR and Azure Bing API:
https://docs.microsoft.com/en-us/azure/cognitive-services/bing-entities-search/use-display-requirements

Project Kinect for Azure:
https://azure.microsoft.com/en-us/campaigns/kinect/

Qualcomm + Microsoft partnership:
https://azure.microsoft.com/en-us/blog/accelerating-ai-on-the-intelligent-edge-microsoft-and-qualcomm-create-vision-ai-developer-kit/

About Randolph

Craft Expert Randolph Widjaja is part of the Cloud team within Craft, the development programme for IT professionals (powered by Centric). If you would like to follow his blog, sign up for Craft updates

Want to know more about Craft, the development programme for IT professionals? Check out the website!

Tags:Cloud

     
Schrijf een reactie
  • Captcha image
  • Verzenden