Skip to content

azure-cognitive-nodered.md

Latest commit

 

History

History
114 lines (74 loc) · 3.85 KB

azure-cognitive-nodered.md

File metadata and controls

114 lines (74 loc) · 3.85 KB

Azure Cognitive services with NodeRED

now that you have the IBM Watson Cogntive services working in NodeRED, let's see what would be needed to do the same/similar with Microsot Azure Cognitive Services.

You will need to add a Computer Vision service subscription to your Azure account. We will use a limited free service subscription for this.

Azure Computer Vision subscription

Let's start with create the computer vision resource in the Azure portal

azure portal - cv

Select "Computer Vision"

azure cv create

Click Create

azure cv settings

Note that the Location property has been overidden to "West US" - this is necessary for the current version of the Azure NodeRED nodes, which do not yet provide a means of selecting different regional endpoints.

Note the Pricing tier has been set to F0 - a free, but restricted plan.

Again, click Create

Once the subscription is active, you will be able to see it in the "All resources" view:

azure all resources

Click on the subscription name (in this case, "brs-cv-westus") to view its details:

azure cv details

At this point, we are particularly interested in the Grab your keys section - the API will be added into the configuration of the NodeRED Azure node. Click onKeys

azure vc keys

Here, you wil see 2 keys - either one can be used -- copy one to your favourite scratchpad for later use.

Use a similar process to set up subscriptions for other Cognitive Services.

NodeRED setup for Azure Cognitive Services

With a minor modification to the existing image classification flow, you can enhance the image interpretation - azure cv flow

For this to work, you will need to install the node-red-contrib-cognitive-services package. Use the "Manage Palette" method to does this. azure install in nodered

You should now have, in the left-hand palette, a new set of nodes

azure nodes

Pull the Computer Vision node onto the canvas and link it into the existing flow, between the Watson Visual Recognition node, and the output template node.

azure cv insert

and configure with the subscription API key

azure cv config.

Almost ready to update the runtime, but the content/format of the output from the Azure service is not known; add a debug node after the computer vison node, to allow inspection of the data returned from the API call.

azure cv debug

Click the Deploy button

deploy

Now, redo the image classification, and observe the output in the NodeRED debug window. The key piece of content that Azure provides is a descriptionn of the image, which you can find in the msg.detail.description.captions array

azure cv response

All that is needed now is to incorporate that information into the response template

azure cv template

The new template will be

{#detail}}
 {{#description}}
  {{#captions}}
   <strong>{{text}} ({{confidence}})</strong>
  {{/captions}}
 {{/description}}
{{/detail}}

{{#result}}
<table>
 {{#images}}
  {{#classifiers}}
    {{#classes}}
        <tr><td>{{class}}</td><td>{{score}}</td><td>{{{type_hierarchy}}}</td></tr>
    {{/classes}}
  {{/classifiers}}
 {{/images}}
</table>
{{/result}}

You can copy/paste this into the template node, if you like.

After deploying again, the user will see extra information on the classification page:

azure cv add info