Microsoft Face


The microsoft_face component platform is the main component for Microsoft Azure Cognitive service Face. All data are stored in your own private instance in the Azure cloud.

You need an API key which is free but requires a Azure registration using your microsoft ID. The free resource (F0) is limited to 20 requests per minute and 30k requests in a month. If you don’t want to use the Azure cloud, you can also get an API key by registering with cognitive-services. However, all keys on cognitive services must be recreated every 90 days.

To enable the Microsoft Face component, add the following lines to your configuration.yaml:

# Example configuration.yaml entry
microsoft_face:
  api_key: YOUR_API_KEY
  azure_region: eastus2

Configuration variables:

  • api_key (Required): The API key for your Cognitive resource.
  • azure_region (Optional): The region where you instantiated your Microsoft Cognitive services endpoint
  • timeout (Optional): Set timeout for the API connection. Defaults to 10s.

Person and Groups

For most services, you need to set up a group or a person. This limits the processing and detection to elements provided by the group. Home Assistent creates an entty for all groups and allows you to show the state, person and IDs directly on the frontend.

The following services are available for managing this feature. They can be called via the Frontend, a script, or the REST API.

  • microsoft_face.create_group
  • microsoft_face.delete_group
service: microsoft_face.create_group
data:
  name: 'Family'
  • microsoft_face.create_person
  • microsoft_face.delete_person
service: microsoft_face.create_person
data:
  group: family
  name: 'Hans Maier'

You need to add an image of a person. You can add multiple images for every person to make the detection better. You can take a picture from a camera or send a local image to your Azure resource.

  • microsoft_face.face_person
service: microsoft_face.face_person
data:
  group: family
  name: 'Hans Maier'
  camera_entity: camera.door

For the local image we need curl. The {personId} is present in group entity as attribute.

$ curl -v -X POST "https://westus.api.cognitive.microsoft.com/face/v1.0/persongroups/{GroupName}/persons/{personId}/persistedFaces" \
  -H "Ocp-Apim-Subscription-Key: YOUR_API_KEY" \
  -H "Content-Type: application/octet-stream" --data-binary "@/tmp/image.jpg"

After we’re done with changes on a group, we need train this group to teach the AI how to handle the new data.

  • microsoft_face.train_group
service: microsoft_face.train_group
data:
  group: family