Face recognition app made with Amazon Rekognition and Java

Introduction

This time, I would like to make a simple application using Amazon's Rekognition face recognition service. When you hear face recognition, you may have a difficult image, but if you use this Rekoginition, you can easily make it, so please try using it through a sample.

Only the excerpted source is posted in the article, so if you are interested, I will post the source I made this time in here. please refer.

Premise

--You have an AWS account. --The access key and secret key of the IAM user who has the policy of "Amazon Rekognition Full Access" must be acquired.

Operating environment

• MacOS Mojave
• Java 8

App screen

This is the screen of the app I made this time.

Development flow

1. Create a collection to store the comparison source image (meta data)
2. Store the comparison source image in the collection
3. Take a picture with the camera at the time of authentication and upload
4. Display the authentication result

Create a collection

Collection creation

ListCollectionsResult listCollectionsResult = rekognitionClient.listCollections(new ListCollectionsRequest());
if (!listCollectionsResult.getCollectionIds().contains(COLLECTION_ID)) {
    rekognitionClient.createCollection(new CreateCollectionRequest().withCollectionId(COLLECTION_ID));
}

Instead of simply creating a collection, I try to create one if it doesn't exist.

Store the comparison source image in the collection

Store comparison source image

IndexFacesRequest indexFacesRequest = new IndexFacesRequest()
              .withCollectionId(COLLECTION_ID)
              .withExternalImageId(id)
              .withImage(new Image().withBytes(ByteBuffer.wrap(multipartFile.getBytes())));

IndexFacesResult indexFacesResult = rekognitionClient.indexFaces(indexFacesRequest);

This time, we will manage the ID that uniquely identifies the user in externalImageId.

Take a picture with the camera at the time of authentication and upload

SearchFacesByImageRequest searchFacesByImageRequest = new SearchFacesByImageRequest()
          .withCollectionId(COLLECTION_ID)
          .withImage(new Image().withBytes(ByteBuffer.wrap(multipartFile.getBytes())))
          .withFaceMatchThreshold(90F)
          .withMaxFaces(1);

try {
    SearchFacesByImageResult searchFacesByImageResult = rekognitionClient.searchFacesByImage(searchFacesByImageRequest);
} catch (InvalidParameterException e) {
    //Error handling
}

Please note that InvalidParameterException will occur if the face cannot be recognized from the uploaded image.

In this sample, we tried to get the one with the highest similarity from the ones with 90% or more similarity.

Display authentication result

SearchFacesByImageResult can be obtained in the following format, so get externalFaceId from it.

{
   "FaceMatches": [ 
      { 
         "Face": { 
            "BoundingBox": { 
               "Height": number,
               "Left": number,
               "Top": number,
               "Width": number
            },
            "Confidence": number,
            "ExternalImageId": "string",  <--this
            "FaceId": "string",
            "ImageId": "string"
         },
         "Similarity": number
      }
   ],
   "FaceModelVersion": "string",
   "SearchedFaceBoundingBox": { 
      "Height": number,
      "Left": number,
      "Top": number,
      "Width": number
   },
   "SearchedFaceConfidence": number
}

If it matches the ID entered on the screen, authentication is OK, otherwise it is NG.

Reference site

-I tried face recognition with Amazon Rekognition