Amazon Rekognition
Developer Guide

IndexFaces

Detects faces in the input image and adds them to the specified collection.

Amazon Rekognition doesn't save the actual faces that are detected. Instead, the underlying detection algorithm first detects the faces in the input image. For each face, the algorithm extracts facial features into a feature vector, and stores it in the backend database. Amazon Rekognition uses feature vectors when it performs face match and search operations using the SearchFaces and SearchFacesByImage operations.

For more information, see Adding Faces to a Collection.

To get the number of faces in a collection, call DescribeCollection.

If you're using version 1.0 of the face detection model, IndexFaces indexes the 15 largest faces in the input image. Later versions of the face detection model index the 100 largest faces in the input image. To determine which version of the model you're using, call DescribeCollection and supply the collection ID. You can also get the model version from the value of FaceModelVersion in the response from IndexFaces.

For more information, see Model Versioning.

If you provide the optional ExternalImageID for the input image you provided, Amazon Rekognition associates this ID with all faces that it detects. When you call the ListFaces operation, the response returns the external ID. You can use this external image ID to create a client-side index to associate the faces with each image. You can then use the index to find all faces in an image.

You can specify the maximum number of faces to index with the MaxFaces input parameter. This is useful when you want to index the largest faces in an image and don't want to index smaller faces, such as those belonging to people standing in the background.

The QualityFilter input parameter allows you to filter out detected faces that don’t meet the required quality bar chosen by Amazon Rekognition. The quality bar is based on a variety of common use cases. By default, IndexFaces filters detected faces. You can also explicitly filter detected faces by specifying AUTO for the value of QualityFilter. If you do not want to filter detected faces, specify NONE.

Note

To use quality filtering, you need a collection associated with version 3 of the face model. To get the version of the face model associated with a collection, call DescribeCollection.

Information about faces detected in an image, but not indexed, is returned in an array of UnindexedFace objects, UnindexedFaces. Faces aren't indexed for reasons such as:

  • The number of faces detected exceeds the value of the MaxFaces request parameter.

  • The face is too small compared to the image dimensions.

  • The face is too blurry.

  • The image is too dark.

  • The face has an extreme pose.

In response, the IndexFaces operation returns an array of metadata for all detected faces, FaceRecords. This includes:

  • The bounding box, BoundingBox, of the detected face.

  • A confidence value, Confidence, which indicates the confidence that the bounding box contains a face.

  • A face ID, faceId, assigned by the service for each face that's detected and stored.

  • An image ID, ImageId, assigned by the service for the input image.

If you request all facial attributes (by using the detectionAttributes parameter), Amazon Rekognition returns detailed facial attributes, such as facial landmarks (for example, location of eye and mouth) and other facial attributes like gender. If you provide the same image, specify the same collection, and use the same external ID in the IndexFaces operation, Amazon Rekognition doesn't save duplicate face metadata.

The input image is passed either as base64-encoded image bytes, or as a reference to an image in an Amazon S3 bucket. If you use the AWS CLI to call Amazon Rekognition operations, passing image bytes isn't supported. The image must be formatted as a PNG or JPEG file.

This operation requires permissions to perform the rekognition:IndexFaces action.

Request Syntax

{ "CollectionId": "string", "DetectionAttributes": [ "string" ], "ExternalImageId": "string", "Image": { "Bytes": blob, "S3Object": { "Bucket": "string", "Name": "string", "Version": "string" } }, "MaxFaces": number, "QualityFilter": "string" }

Request Parameters

The request accepts the following data in JSON format.

CollectionId

The ID of an existing collection to which you want to add the faces that are detected in the input images.

Type: String

Length Constraints: Minimum length of 1. Maximum length of 255.

Pattern: [a-zA-Z0-9_.\-]+

Required: Yes

DetectionAttributes

An array of facial attributes that you want to be returned. This can be the default list of attributes or all attributes. If you don't specify a value for Attributes or if you specify ["DEFAULT"], the API returns the following subset of facial attributes: BoundingBox, Confidence, Pose, Quality, and Landmarks. If you provide ["ALL"], all facial attributes are returned, but the operation takes longer to complete.

If you provide both, ["ALL", "DEFAULT"], the service uses a logical AND operator to determine which attributes to return (in this case, all attributes).

Type: Array of strings

Valid Values: DEFAULT | ALL

Required: No

ExternalImageId

The ID you want to assign to all the faces detected in the image.

Type: String

Length Constraints: Minimum length of 1. Maximum length of 255.

Pattern: [a-zA-Z0-9_.\-:]+

Required: No

Image

The input image as base64-encoded bytes or an S3 object. If you use the AWS CLI to call Amazon Rekognition operations, passing base64-encoded image bytes isn't supported.

Type: Image object

Required: Yes

MaxFaces

The maximum number of faces to index. The value of MaxFaces must be greater than or equal to 1. IndexFaces returns no more than 100 detected faces in an image, even if you specify a larger value for MaxFaces.

If IndexFaces detects more faces than the value of MaxFaces, the faces with the lowest quality are filtered out first. If there are still more faces than the value of MaxFaces, the faces with the smallest bounding boxes are filtered out (up to the number that's needed to satisfy the value of MaxFaces). Information about the unindexed faces is available in the UnindexedFaces array.

The faces that are returned by IndexFaces are sorted by the largest face bounding box size to the smallest size, in descending order.

MaxFaces can be used with a collection associated with any version of the face model.

Type: Integer

Valid Range: Minimum value of 1.

Required: No

QualityFilter

A filter that specifies how much filtering is done to identify faces that are detected with low quality. Filtered faces aren't indexed. If you specify AUTO, filtering prioritizes the identification of faces that don’t meet the required quality bar chosen by Amazon Rekognition. The quality bar is based on a variety of common use cases. Low-quality detections can occur for a number of reasons. Some examples are an object that's misidentified as a face, a face that's too blurry, or a face with a pose that's too extreme to use. If you specify NONE, no filtering is performed. The default value is AUTO.

To use quality filtering, the collection you are using must be associated with version 3 of the face model.

Type: String

Valid Values: NONE | AUTO

Required: No

Response Syntax

{ "FaceModelVersion": "string", "FaceRecords": [ { "Face": { "BoundingBox": { "Height": number, "Left": number, "Top": number, "Width": number }, "Confidence": number, "ExternalImageId": "string", "FaceId": "string", "ImageId": "string" }, "FaceDetail": { "AgeRange": { "High": number, "Low": number }, "Beard": { "Confidence": number, "Value": boolean }, "BoundingBox": { "Height": number, "Left": number, "Top": number, "Width": number }, "Confidence": number, "Emotions": [ { "Confidence": number, "Type": "string" } ], "Eyeglasses": { "Confidence": number, "Value": boolean }, "EyesOpen": { "Confidence": number, "Value": boolean }, "Gender": { "Confidence": number, "Value": "string" }, "Landmarks": [ { "Type": "string", "X": number, "Y": number } ], "MouthOpen": { "Confidence": number, "Value": boolean }, "Mustache": { "Confidence": number, "Value": boolean }, "Pose": { "Pitch": number, "Roll": number, "Yaw": number }, "Quality": { "Brightness": number, "Sharpness": number }, "Smile": { "Confidence": number, "Value": boolean }, "Sunglasses": { "Confidence": number, "Value": boolean } } } ], "OrientationCorrection": "string", "UnindexedFaces": [ { "FaceDetail": { "AgeRange": { "High": number, "Low": number }, "Beard": { "Confidence": number, "Value": boolean }, "BoundingBox": { "Height": number, "Left": number, "Top": number, "Width": number }, "Confidence": number, "Emotions": [ { "Confidence": number, "Type": "string" } ], "Eyeglasses": { "Confidence": number, "Value": boolean }, "EyesOpen": { "Confidence": number, "Value": boolean }, "Gender": { "Confidence": number, "Value": "string" }, "Landmarks": [ { "Type": "string", "X": number, "Y": number } ], "MouthOpen": { "Confidence": number, "Value": boolean }, "Mustache": { "Confidence": number, "Value": boolean }, "Pose": { "Pitch": number, "Roll": number, "Yaw": number }, "Quality": { "Brightness": number, "Sharpness": number }, "Smile": { "Confidence": number, "Value": boolean }, "Sunglasses": { "Confidence": number, "Value": boolean } }, "Reasons": [ "string" ] } ] }

Response Elements

If the action is successful, the service sends back an HTTP 200 response.

The following data is returned in JSON format by the service.

FaceModelVersion

The version number of the face detection model that's associated with the input collection (CollectionId).

Type: String

FaceRecords

An array of faces detected and added to the collection. For more information, see Managing Faces in a Collection.

Type: Array of FaceRecord objects

OrientationCorrection

The orientation of the input image (counterclockwise direction). If your application displays the image, you can use this value to correct image orientation. The bounding box coordinates returned in FaceRecords represent face locations before the image orientation is corrected.

Note

If the input image is in jpeg format, it might contain exchangeable image (Exif) metadata. If so, and the Exif metadata populates the orientation field, the value of OrientationCorrection is null. The bounding box coordinates in FaceRecords represent face locations after Exif metadata is used to correct the image orientation. Images in .png format don't contain Exif metadata.

Type: String

Valid Values: ROTATE_0 | ROTATE_90 | ROTATE_180 | ROTATE_270

UnindexedFaces

An array of faces that were detected in the image but weren't indexed. They weren't indexed because the quality filter identified them as low quality, or the MaxFaces request parameter filtered them out. To use the quality filter, you specify the QualityFilter request parameter.

Type: Array of UnindexedFace objects

Errors

AccessDeniedException

You are not authorized to perform the action.

HTTP Status Code: 400

ImageTooLargeException

The input image size exceeds the allowed limit. For more information, see Limits in Amazon Rekognition.

HTTP Status Code: 400

InternalServerError

Amazon Rekognition experienced a service issue. Try your call again.

HTTP Status Code: 500

InvalidImageFormatException

The provided image format is not supported.

HTTP Status Code: 400

InvalidParameterException

Input parameter violated a constraint. Validate your parameter before calling the API operation again.

HTTP Status Code: 400

InvalidS3ObjectException

Amazon Rekognition is unable to access the S3 object specified in the request.

HTTP Status Code: 400

ProvisionedThroughputExceededException

The number of requests exceeded your throughput limit. If you want to increase this limit, contact Amazon Rekognition.

HTTP Status Code: 400

ResourceNotFoundException

The collection specified in the request cannot be found.

HTTP Status Code: 400

ThrottlingException

Amazon Rekognition is temporarily unable to process the request. Try your call again.

HTTP Status Code: 500

See Also

For more information about using this API in one of the language-specific AWS SDKs, see the following: