Granted Patent B2

US 12,417,561 · App. 18/128,906 · Granted Sep 16, 2025

Inventors: Michele Saad (Austin, TX); Ajay Jain (San Jose, CA)

Assignee: Adobe Inc.

G06T11/00G06N3/045G06N3/0475G06N3/094G06Q30/0631G06Q50/01

Abstract

Claims (74)

1. A computer-implemented method comprising:

receiving, by an image generation module, one or more parameters from a client device, the one or more parameters characterizing a subject displayed on the client device;

responsive to receiving the one or more parameters, generating, by the image generation module, a parameterized image of the subject using a generative machine learning model, wherein the one or more parameters are used as a seed by the generative machine learning model for generating the parameterized image of the subject;

outputting, by the image generation module, the parameterized image of the subject for display to the client device;

generating, by a recommendation module, a first set of recommendations for a user of the client device, the first set of recommendations based on at least the subject;

receiving, by the recommendation module, the one or more parameters;

determining, by the recommendation module, based on the one or more parameters and the subject, a second set of recommendations for the user of the client device; and

outputting, by the recommendation module, the second set of recommendations to the client device.

2. The method of claim 1 , wherein the one or more parameters includes a parameter selected by the user of the client device.

3. The method of claim 1 , wherein the one or more parameters includes a parameter generated by a machine learning model and inferred from a behavior of the user of the client device.

4. The method of claim 3 , wherein the second set of recommendations is determined using at least one of a behavioral segmentation algorithm, a content-based filtering algorithm, or a collaborative filtering algorithm.

5. The method of claim 3 , wherein the first set of recommendations for the user of the client device are based on the behavior of the user of the client device.

6. The method of claim 3 , wherein the first set of recommendations is ranked by the recommendation module; and

wherein determining the second set of recommendations comprises re-ranking, by the recommendation module, the first set of recommendations based on the one or more parameters.

7. The method of claim 1 , wherein the one or more parameters includes a first parameter selected by the user of the client device and a second parameter generated by a machine learning model and inferred from a behavior of the user of the client device.

8. The method of claim 1 , further comprising:

determining, by a behavior analysis module, from a behavior of the user of the client device, a behavioral metric of the user; and

inferring, by the behavior analysis module, from the behavioral metric, an inferred parameter; and

wherein the image generation module uses the inferred parameter as another seed for the generative machine learning model to generate the parameterized image.

9. The method of claim 1 , further comprising:

determining, by a behavior analysis module, from a behavior of a first user of a first client device, a first behavioral metric;

determining, by the behavior analysis module, from a behavior of a second user of a second client device, a second behavioral metric;

generating, by the behavior analysis module, using the first behavioral metric and the second behavioral metric, a similarity measure between the first user and the second user;

determining, by the behavior analysis module, from the similarity measure, one or more parameters for the second user;

generating, by the image generation module, an inferred parameterized image using the generative machine learning model, wherein the one or more parameters for the second user are used as a second seed for the generative machine learning model to generate the inferred parameterized image; and

outputting, by the image generation module, the inferred parameterized image for display to the second client device.

10. The method of claim 1 , further comprising:

determining, by a behavior analysis module, from a first behavior of the user of the client device, a first behavioral metric;

determining, by the behavior analysis module, from a second behavior of the user of the client device, a second behavioral metric;

determining, by the behavior analysis module and using a reinforcement learning model, a behavior pattern from the first behavioral metric and the second behavioral metric;

determining, by the behavior analysis module, from the behavior pattern, one or more parameters for the user;

generating, by the image generation module, an inferred parameterized image using the generative machine learning model, wherein the one or more parameters for the user are used as a second seed for the generative machine learning model to generate the inferred parameterized image; and

outputting, by the image generation module, the inferred parameterized image to the client device.

11. A synthetic image generation system comprising:

an image generation module configured to:

receive one or more parameters from a client device, the one or more parameters characterizing a subject displayed on the client device;

responsive to receive the one or more parameters, generate a parameterized image of the subject using a generative machine learning model, wherein the one or more parameters are used as a seed by the generative machine learning model for generating the parameterized image of the subject; and

output the parameterized image of the subject for display to the client device; and

a recommendation module configured to:

generate a first set of recommendations for a user of the client device, the first set of recommendations based on at least the subject;

receive the one or more parameters;

determine, based on the one or more parameters and the subject, a second set of recommendations for the user of the client device; and

output the second set of recommendations to the client device.

12. The synthetic image generation system of claim 11 , wherein the one or more parameters includes a parameter selected by the user of the client device.

13. The synthetic image generation system of claim 11 , wherein the one or more parameters includes a parameter generated by a machine learning model and inferred from a behavior of the user of the client device.

14. The synthetic image generation system of claim 11 , wherein the one or more parameters includes a first parameter selected by the user of the client device and a second parameter generated by a machine learning model and inferred from a behavior of the user of the client device.

15. The synthetic image generation system of claim 11 , wherein the synthetic image generation system further comprises a behavior analysis module configured to:

determine from a behavior of the user of the client device, a behavioral metric of the user; and

infer, from the behavioral metric, an inferred parameter; and

wherein the image generation module uses the inferred parameter as another seed for the generative machine learning model to generate the parameterized image.

16. The synthetic image generation system of claim 11 , wherein the synthetic image generation system further comprises a behavior analysis module configured to:

determine, from a behavior of a first user of a first client device, a first behavioral metric;

determine, from a behavior of a second user of a second client device, a second behavioral metric;

generate, using the first behavioral metric and the second behavioral metric, a similarity measure between the first user and the second user;

determine, from the similarity measure, one or more parameters for the second user; and wherein the image generation module is further configured to:

generate an inferred parameterized image using the generative machine learning model, wherein the one or more parameters for the second user are used as a second seed for the generative machine learning model to generate the inferred parameterized image; and

output the inferred parameterized image to the second client device.

17. A non-transitory computer-readable storage medium including instructions configured to cause one or more processing devices to perform actions including:

receiving one or more parameters from a client device, the one or more parameters characterizing a subject displayed on the client device;

responsive to receive the one or more parameters, generating a parameterized image of the subject using a generative machine learning model, wherein the one or more parameters are used as a seed by the generative machine learning model for generating the parameterized image of the subject;

outputting the parameterized image of the subject for display to the client device;

generating a first set of recommendations for a user of the client device, the first set of recommendations based on at least the subject;

receiving the one or more parameters;

determining, based on the one or more parameters, a second set of recommendations for the user of the client device and the subject; and

outputting the second set of recommendations to the client device.

18. The non-transitory computer-readable medium of claim 17 , wherein the one or more parameters includes a parameter selected by the user of the client device.

19. The non-transitory computer-readable medium of claim 17 , wherein the one or more parameters includes a parameter generated by a machine learning model and inferred from a behavior of the user of the client device.

20. The non-transitory computer-readable medium of claim 17 , further including the actions:

determining from a behavior of a first user of a first client device, a first behavioral metric;

determining from a behavior of a second user of a second client device, a second behavioral metric;

generating using the first behavioral metric and the second behavioral metric, a similarity measure between the first user and the second user;

determining from the similarity measure, one or more parameters for the second user;

generating an inferred parameterized image using the generative machine learning model, wherein the one or more parameters for the second user are used as a second seed for the generative machine learning model to generate the inferred parameterized image; and

outputting the inferred parameterized image to the second client device.

Assignments (1)

Continuity (1)

References Cited (13)