Is Amazon Rekognition able to detect fake photos on identity documents?

0

Hello, I have a question regarding Amazon Rekognition: Can it detect fake photos on identity documents? We're currently working on integrating an onboarding solution into our system. We're concerned about the possibility of fake users uploading others' identity documents and manipulating them with photo editing software to match their faces. Is Amazon Rekognition capable of detecting this scenario? Your assistance would be greatly appreciated. Thank you.

gefragt vor einem Monat157 Aufrufe
1 Antwort
1

On its own it does not have the capability to verify if documents have been tampered with or manipulated . However Rekognition offers some features that can help mitigate the risk of fake photos on identity documents:

1/ Face Liveness: This feature helps detect spoofs like printed photos, digital images, or masks presented to the camera. It can improve the chances of catching a fake document being used in place of a real person.

2/ Face Comparison: You can compare the selfie image of the user with the face in the identity document. This helps ensure they match and reduces the possibility of a different person's ID being used. Or maybe , use the Face Collection feature to take facial images from documents and match them against a collection of known user profiles. Face search APIs can be used for this matching.

3/Implement human review workflows using Sagemaker Ground Truth or Augmented AI to validate any low confidence matches from the above steps. Humans can cross-check for any anomalies or tampering not detected technically.

For more robust onboarding process, it's wise to use Rekognition alongside other security measures.

profile pictureAWS
akad
beantwortet vor einem Monat
profile pictureAWS
EXPERTE
iBehr
überprüft vor einem Monat

Du bist nicht angemeldet. Anmelden um eine Antwort zu veröffentlichen.

Eine gute Antwort beantwortet die Frage klar, gibt konstruktives Feedback und fördert die berufliche Weiterentwicklung des Fragenstellers.

Richtlinien für die Beantwortung von Fragen