Granted Patent B2

US 12,670,297 · App. 18/480,690 · Granted Jun 30, 2026

Inventor: Daniil Baryshnikov (London, GB)

Assignee: BUMBLE IP HOLDCO LLC

G06F21/84G06F21/6263G06T5/70G06V10/70G06V40/168G06V40/172G06T2207/20081G06T2207/20212G06T2207/30201

Abstract

Claims (65)

1 . A computer-implemented method comprising:

obtaining, from a first client device associated with a first user, digital image data, wherein the first user is one of a plurality of users of a social network;

determining that the digital image data includes an image of multiple unique faces;

applying, to the digital image data, a machine learning model configured to identify, among the multiple unique faces, a face of the first user;

selecting an obscuring mechanism that specifies a type of action to obscure image regions in the digital image data corresponding to the multiple unique faces except the face of the first user based on a preference of the first user;

obscuring, in the digital image data and using the type of action specified by the selected obscured mechanism based on the preference of the first user, the multiple unique faces except the face of the first user on the image regions;

storing the digital image data having the obscured faces in a database;

associating the stored digital image data with a profile associated with the first user in the social network;

obtaining, from the first client device, a plurality of digital image data that include an image of the face of the first user; and

training, based on the plurality of digital image data, the machine learning model configured to extract features representative of the face of the first user.

2 . The computer-implemented method of claim 1 , wherein the digital image data comprise (i) one or more pictures, (ii) one or more videos, or both.

3 . The computer-implemented method of claim 1 , further comprising:

providing, to the first client device and at least some of a plurality of client devices associated with users of the social network, the digital image data.

4 . The computer-implemented method of claim 1 , further comprising:

prompting the first user to take a live selfie; and

verifying an identity of the first user based on the live selfie.

5 . The computer-implemented method of claim 1 , further comprising:

in response to obtaining, from the first client device, feedback data specifying additional faces to be obscured, obscuring the additional faces in the digital image data.

6 . The computer-implemented method of claim 1 , wherein the obscuring mechanism comprises one or more of (i) blurring a face in the image regions in the digital image data and (ii) overlaying a visual representation on pixels indicative of the face in the image regions in the digital image data.

7 . The computer-implemented method of claim 6 , wherein the visual representation comprises one or more of an emoji and a face of the first user.

8 . The computer-implemented method of claim 1 , further comprising:

obtaining, from a second client device associated with a second user, a privacy preference specifying approval to display a face of the second user in the digital image data associated with the profile of the first user; and

obscuring the multiple unique faces in the digital image data except the face of the first user and the face of the second user.

9 . The computer-implemented method of claim 1 , wherein the users of the social network are users matched to the first user on the social network.

10 . The computer-implemented method of claim 1 , wherein each of the digital image data is accompanied with a text caption.

11 . The computer-implemented method of claim 1 , further comprising:

determining that the first user violates a privacy requirement of the social network; and

presenting, to the first client device, a warning indication.

12 . The computer-implemented method of claim 11 , wherein determining that the first user violates the privacy requirement of the social network comprises:

obtaining an indication, from a third client device associated with a third user matched to the first user, that at least one of the digital image data in the profile associated with the first user includes a second face other than the face of the first user; and

determining that the second face does not belong to users who approved usage of their faces to the first user.

13 . The computer-implemented method of claim 1 , further comprising:

applying, to the digital image data, a second machine learning model configured to identify a portion of the digital image data that do not meet one or more safety criteria; and

obscuring, in the digital image data, the portion of the digital image data.

14 . The computer-implemented method of claim 13 , wherein the portion of the digital image data that do not meet the safety criteria comprises age-inappropriate contents.

15 . A system comprising:

one or more processors and one or more storage devices storing instructions that are operable, when executed by the one or more processors, to cause the one or more processors to perform operations comprising:

obtaining, from a first client device associated with a first user, digital image data, wherein the first user is one of a plurality of users of a social network;

determining that the digital image data includes an image of multiple unique faces;

applying, to the digital image data, a machine learning model configured to identify, among the multiple unique faces, a face of the first user;

selecting an obscuring mechanism that specifies a type of action to obscure image regions in the digital image data corresponding to the multiple unique faces except the face of the first user based on a preference of the first user;

obscuring, in the digital image data and using the type of action specified by the selected obscured mechanism based on the preference of the first user, the multiple unique faces except the face of the first user on the image regions;

storing the digital image data having the obscured faces in a database;

associating the stored digital image data with a profile associated with the first user in the social network;

obtaining, from the first client device, a plurality of digital image data that include an image of the face of the first user; and

training, based on the plurality of digital image data, the machine learning model configured to extract features representative of the face of the first user.

16 . The system of claim 15 , further comprising:

prompting the first user to take a live selfie; and

verifying an identity of the first user based on the live selfie.

17 . The system of claim 15 , further comprising:

obtaining, from a second client device associated with a second user, a privacy preference specifying approval to display a face of the second user in the digital image data associated with the profile of the first user; and

obscuring the multiple unique faces in the digital image data except the face of the first user and the face of the second user.

18 . The system of claim 15 , further comprising:

applying, to the digital image data, a second machine learning model configured to identify a portion of the digital image data that do not meet one or more safety criteria; and

obscuring, in the digital image data, the portion of the digital image data.

19 . A non-transitory computer-readable medium, comprising software instructions, that when executed by a computer, cause the computer to execute operations comprising:

obtaining, from a first client device associated with a first user, digital image data, wherein the first user is one of a plurality of users of a social network;

determining that the digital image data includes an image of multiple unique faces;

applying, to the digital image data, a machine learning model configured to identify, among the multiple unique faces, a face of the first user;

selecting an obscuring mechanism that specifies a type of action to obscure image regions in the digital image data corresponding to the multiple unique faces except the face of the first user based on a preference of the first user;

obscuring, in the digital image data and using the type of action specified by the selected obscured mechanism based on the preference of the first user, the multiple unique faces except the face of the first user on the image regions;

storing the digital image data having the obscured faces in a database;

associating the stored digital image data with a profile associated with the first user in the social network;

obtaining, from the first client device, a plurality of digital image data that include an image of the face of the first user; and

training, based on the plurality of digital image data, the machine learning model configured to extract features representative of the face of the first user.

Assignments (5)

Continuity (2)

References Cited (125)