Im trying to build a system where users uploads a selfie, and then when someone upload image to some gallery in the ui.
Every new upload should trigger a faces checks against the faces that already uploaded to the collection inside rekognition, and I wanna be able to know if there are matches or no.
Until know I thought to index a face into collection and that grab the Face ID and save it inside user record.
The next step supposed to be to take image as input, index it too and get an array of faces ids present in the picture. Unfortunately, even though I upload myself with others in the picture even though my selfie pic is in the collection, I get different Face ID so I can’t rely on that or cross it with the db.
The question is: The only solution is to operate search operation inside a collection base of an image input?
If so, it’s means that if I upload 100 selfies and 100 input images
I will have to scan and pay for 100x100 scans?
I tried to cross the images on base Face ID but it’s unreliable
You use
IndexFaces()to store a face in a Face Collection. It can accept multiple faces, but it is best to give it just one face so you know which face is linked to which FaceId.Later, when you want to check whether a face is already in a Face Collection you can pass an image to
SearchFacesByImage()that "first detects the largest face in the image, and then searches the specified collection for matching faces." If the face is found, it will returning the FaceId of the previously-stored face.Each of these operations cost
$0.001.