Skip to main content

Amazon Rekognition

See Amazon Rekognition Developer Guide for general guidance using Rekognition.

API Access​

The IRekognitionClient interface and its implementation TRekognitionClient provides access to Rekognition operations.

program DetectFacesInVideo;

uses
  AWS.Rekognition;

var
  Client: IRekognitionClient;
  Request: IRekognitionStartFaceDetectionRequest;
  Response: IRekognitionStartFaceDetectionResponse;

begin
  Client := TRekognitionClient.Create;

  Request := TRekognitionStartFaceDetectionRequest.Create(
    TRekognitionVideo.FromS3Object('my-input-bucket', 'video.mp4')
  );
  try
    Response := Client.StartFaceDetection(Request);
    // Inspect Response which includes a JobId to later retrieve results.
  except
    on E: ERekognitionException do
      Writeln(E.ClassName, ': ', E.Message);
  end;
end.

Detecting objects and concepts​

Labels (objects and concepts) can be detected in images using DetectLabels.

Labels can be detected in stored video (on Amazon S3) by starting an asynchronous job with StartLabelDetection, and later retrieving the results with GetLabelDetection.

Amazon Rekognition Custom Labels can be trained to identify objects, logos, and scenes specific to your business needs.

Detecting and analyzing faces​

Faces can be detected in images using DetectFaces.

Faces can be detected in stored video (on Amazon S3) by starting an asynchronous job with StartFaceDetection, and later retrieving the results with GetFaceDetection.

Faces can also be compared using CompareFaces.

Searching faces in a collection​

Faces can be stored in a collection for searching images or videos.

First create a collection with CreateCollection, and then index faces into it using IndexFaces.

Once indexed, the collection can be used in SearchFaces, and SearchFacesByImage.

Multiple faces can be aggregated to improve identification of people. First use CreateUser to create a "user", and then AssociateFaces with the user. Users can then be searched for within an image using SearchUsers, or SearchUsersByImage.

To search for faces within a stored video (on Amazon S3), you can use the StartFaceSearch function, and later retrieve the results using GetFaceSearch.

Detecting personal protective equipment​

Personal Protective Equipment (PPE) can be detected in an image using DetectProtectiveEquipment.

Recognizing celebrities​

Amazon Rekognition is capable of recognizing tens of thousands of well-known personalities.

Detect celebrities in images using RecognizeCelebrities. Celebrities can be recognized in stored video (on Amazon S3) using StartCelebrityRecognition and later retrieving the results with GetCelebrityRecognition.

Moderating content​

Amazon Rekognition can be used to detect content that is inappropriate, unwanted, or offensive.

Content in images can be labeled using DetectModerationLabels. Content in stored video (on Amazon S3) can be detected using StartContentModeration, and later retrieving the results with GetContentModeration.

Detecting text​

Text can be detected in images using DetectText.

Text in stored video (on Amazon S3) can be detected using StartTextDetection, and later retrieving the results with GetTextDetection.

Detecting video segments​

Segments in stored video (on Amazon S3) can be detected such as black frames and end credits. Use StartSegmentDetection to detect segments, and later retrieve the results with GetSegmentDetection.

Bulk analysis​

Bulk analysis of images can be performed providing an input manifest to StartMediaAnalysisJob, and later retrieving the results manifest through GetMediaAnalysisJob.