Authentication of age, gender, and other biometric data from live images of users

Patent metadata

Field	Value
Publication number	US-12243336-B2
Application number	US-202217708132-A
Country	US
Kind code	B2
Filing date	Mar 30, 2022
Priority date	Mar 30, 2022
Publication date	Mar 4, 2025
Grant date	Mar 4, 2025

Abstract

A system and method is disclosed for improving fraud detection in the context of a user submitting, via a client device, a photo of a photo ID and a selfie taken during a step of the verification process. A machine learning model may be trained to generate biometric data signals from the selfie. As examples, the biometric data signals generated from the selfie can include an estimated age, gender, hair color, and eye color. The biometric data generated from the selfie may be compared with biometric data listed on the photo ID as an aid to identifying potential identity fraud. In some implementations, a facial filter corresponding to a map of a set of facial measurements of the photo in the photo ID may be compared with the facial measurements of the selfie.

First claim

What is claimed is:

1. A computer-implemented method for user authentication comprising:
receiving a photo of a photo identification document of a human being to be authenticated;
receiving at least one additional photo of the human being to be authenticated taken live during a verification test;
identifying first biometric data listed in the photo identification document, including listed age data;
generating second biometric data based on the photo of the photo identification document utilizing a machine learning model trained to identify biometric age data from photos, the second biometric data including a first estimated age and an aging rate with respect to the listed age data of the photo identification document;
generating third biometric data based on the at least one additional photo by analyzing the at least one additional photo utilizing the machine learning model trained to identify biometric data from photos including generating a second estimated age;
calibrating the second estimated age based on the aging rate; and
generating a fraud signal based on a comparison of the first biometric data listed in the photo identification document and the generated second and third biometric data.

2. The computer-implemented method of claim 1 , wherein generating the third biometric data based on the at least one additional photo comprises generating at least one of an age and a gender based on the at least one additional photo.

3. The computer-implemented method of claim 2 , wherein generating the fraud signal comprises comparing a least one of an age and a gender between the first biometric data of the photo identification document, the second biometric data, and the third biometric data generated based on the at least one additional photo.

4. The computer-implemented method of claim 2 , wherein the biometric data listed in the photo identification document further comprises an eye color and a hair color.

5. The computer-implemented method of claim 4 , wherein generating the fraud signal comprises comparing an age, a gender, an eye color, and a hair color between the biometric data of the photo identification document and the biometric data generated based on the at least one additional photo.

6. The computer-implemented method of claim 1 , further comprising implementing a facial filter to identify a facial feature map of facial features in the photo in the photo identification document and in the at least one additional photo, and generating the fraud signal further comprises comparing the facial feature map of the photo in the photo identification document and in the at least one additional photo.

7. The computer-implemented method of claim 1 , wherein
generating the fraud signal includes comparing the identified first biometric data listed in the photo identification document with second biometric data generated based on the photo of the photo identification document.

8. The computer-implemented method of claim 1 , wherein generating the first and second estimated age is based at least in part on a model of facial skin aging, facial bone aging, and facial muscle aging.

9. The computer-implemented method of claim 1 , wherein generating the third biometric data from the at least one live photo comprises analyzing, at a higher level of resolution, selected patches of a face indicative of aging to reduce a computational load.

10. A computer implemented system, comprising:
a machine learning fraud model trained to analyze biometric data in photos from: 1) a photo of a photo identification document submitted by a user and 2) a set of photos or a video of a face of the user taken at different angles sufficient to generate a three-dimensional model of a face for age estimation and gender estimation, during a verification step;
the machine learning fraud model generating a fraud indicator signal for a fraud detector based at least in part on a comparison of biometric data listed in the photo identification document and biometric data identified by the machine learning fraud model from the set of photos or video of the user taken during the verification step;
wherein the biometric data listed in the photo identification document includes a listed eye color, a listed hair color and the biometric data identified in the photo or video includes an age, a gender, an eye color, and a hair color;
wherein the biometric data generated by the machine learning fraud model includes an age, a gender, an eye color, and a hair color; and
wherein the system compares the biometric data listed in the photo identification document with biometric data generated based on the photo of the photo identification document and determines if there is a mismatch indicative of a fraudulent photo identification document.

Assignees

• Jumio Corp

Inventors

• Bayer Lukas
• Wang Can
• Rizo Rodriguez Dayron
• Taylor Jason Richard

Classifications

• G06V10/751

  Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching · CPC title

• G06V40/168

  Feature extraction; Face representation · CPC title

• G06V10/56

  relating to colour · CPC title

• G06V40/179

  metadata assisted face recognition · CPC title

• G06V40/172 Primary

  Classification, e.g. identification · CPC title