Rekognition actions

The AWS Rekognition extension provides actions for searching and identifying faces in images within Workforce flows. Use these actions to perform facial recognition against a configured Rekognition collection as part of your automated content processes.

You can install this action extension from the Workforce integration marketplace.

Tip

These action descriptions cover the most commonly used inputs and outputs. When designing flows, you may also notice additional input and output variables.

Prerequisites

The Amazon Rekognition extension uses AWS credentials and a Rekognition collection for authentication. Configure the following environment variables before using the action.

Variable	Type	Required	Description
AWS_REGION	string	Yes	The AWS region used by the extension.
AWS_REKOGNITION_KEY	string	Yes	The AWS Rekognition access key. This value is write-only.
AWS_REKOGNITION_SECRET	string	Yes	The AWS Rekognition secret key. This value is write-only.
AWS_REKOGNITION_COLLECTION_ID	string	Yes	The Rekognition collection ID used for face search.

For details of how to obtain these authentication credentials, see the Amazon Rekognition documentation.

Supported AWS regions

The AWS_REGION setting must use one of the regions supported by the extension, for example us-east-1, us-west-2, eu-west-1, eu-west-2, or ap-southeast-1.

Search faces by image

Searches for faces in an image and returns matching faces from the configured Amazon Rekognition collection.

Inputs

Field	Type	Required	Description
imageUrl	string	Yes	URL of the image containing faces to search.

Outputs

The action returns the matching faces found for the supplied image, together with information about the searched face and the face model used.

Output	Type	Description
FaceMatches	array	A list of face match objects. See Face match object below
FaceModelVersion	string	The version of the face model used for the search
SearchedFaceBoundingBox	object	Bounding box of the face used as the search input (Height, Left, Top, Width)
SearchedFaceConfidence	number	Confidence score for the face detected in the input image

Face match object

Each entry in the FaceMatches array contains the following fields:

Field	Type	Description
Similarity	number	Similarity score between the searched face and the matched face
Face.FaceId	string	Unique identifier for the matched face
Face.ImageId	string	Identifier of the image containing the matched face
Face.ExternalImageId	string	User-defined identifier associated with the matched face's image
Face.Confidence	number	Confidence score for the matched face
Face.IndexFacesModelVersion	string	Version of the model used when the face was indexed
Face.BoundingBox	object	Bounding box of the matched face in its source image (Height, Left, Top, Width)

Example output

{
  "FaceMatches": [
    {
      "Face": {
        "BoundingBox": {
          "Height": 0.25,
          "Left": 0.35,
          "Top": 0.2,
          "Width": 0.18
        },
        "Confidence": 99.9,
        "ExternalImageId": "example-person",
        "FaceId": "12345678-1234-1234-1234-123456789012",
        "ImageId": "87654321-4321-4321-4321-210987654321",
        "IndexFacesModelVersion": "7.0"
      },
      "Similarity": 98.5
    }
  ],
  "FaceModelVersion": "7.0",
  "SearchedFaceBoundingBox": {
    "Height": 0.24,
    "Left": 0.34,
    "Top": 0.19,
    "Width": 0.17
  },
  "SearchedFaceConfidence": 99.8
}

Related pages

Adding actions to flows

Inputs, outputs and ports

Input field variables

Developing extensions