Granted Patent B2

US 12,386,892 · App. 18/416,110 · Granted Aug 12, 2025

Inventors: Sahir Nasir (San Jose, CA); Alan Waterman (Merced, CA)

Assignee: ADEIA GUIDES INC.

G06F16/738G06F16/71G06F16/73G06F16/783G06F16/7837G06F16/7867G06F18/214G06F18/24G06N3/04G06N3/08G06V10/764G06V10/82G06V20/42G06V20/47G06V20/48G06V20/49G06V40/1365G06V40/23

Abstract

Claims (50)

1. A method comprising:

identifying, in a first video segment, a first character performing an action, wherein the action comprises a sequence of sub-actions;

accessing a first mapping associated with the first video segment, wherein the first mapping indicates first action attribute values and time periods of the first action attribute values corresponding to the sequence of sub-actions;

inputting the first action attribute values and the time periods of the first action attribute values of each sub-action to a neural network model;

determining, using the neural network model, an action identifier for the sequence of sub-actions;

searching, based on the action identifier, a content database for a plurality of mappings associated with respective video segments;

comparing, using the neural network model, each mapping of the plurality of mappings to the first mapping;

determining, based on the comparing using the neural network model, a second mapping that matches the first mapping, wherein the second mapping is associated with a second video segment depicting a second character performing the action, and wherein the second mapping indicates second action attribute values and time periods of the second action attribute values corresponding to the second character performing the action;

generating, based on the first and the second mappings, a third mapping that indicates the second action attribute values and time periods of the second action attribute values, wherein the third mapping is associated with a video segment depicting the first character performing the action based on the second action attribute values and time periods of the second action attribute values;

retrieving a content structure comprising an object corresponding to the first character, wherein the content structure comprises virtual modelling data corresponding to the first character; and

modifying the virtual modelling data based on the third mapping.

2. The method of claim 1 , further comprising generating, based on the modified virtual modelling data, a reconstructed first character performing the action based on the second action attribute values and time periods of the second action attribute values.

3. A method comprising:

identifying, in a first video segment, a first character performing an action, wherein the action comprises a sequence of sub-actions;

accessing a first mapping associated with the first video segment, wherein the first mapping indicates action attribute values and time periods of the action attribute values corresponding to the sequence of sub-actions;

inputting the action attribute values and the time periods of the action attribute values of each sub-action to a neural network model;

determining, using the neural network model, an action identifier for the sequence of sub-actions;

searching, based on the action identifier, a content database for a plurality of mappings associated with respective video segments;

generating a first fingerprint based on the first mapping;

comparing, using the neural network model, each mapping of the plurality of mappings to the first mapping, wherein comparing the each mapping of the plurality of mappings to the first mapping comprises:

generating a respective fingerprint based on the each mapping; and

comparing the first fingerprint and the respective fingerprint; and

determining, based on the comparing using the neural network model, a second mapping that matches the first mapping, wherein the second mapping is associated with a second video segment depicting a second character performing the action.

4. The method of claim 1 , wherein the action identifier comprises an action keyword based on the sequence of sub-actions.

5. The method of claim 1 , wherein the content database comprises deconstructed video segments depicting expert actions.

6. The method of claim 1 , further comprising generating for display visual representations of the each mapping of the plurality of mappings.

7. The method of claim 6 , wherein the visual representations are displayed on a graphical interface.

8. A system comprising:

one or more input/output (I/O) paths configured to access content; and

control circuitry configured to:

identify, in a first video segment, a first character performing an action, wherein the action comprises a sequence of sub-actions;

access a first mapping associated with the first video segment, wherein the first mapping indicates first action attribute values and time periods of the first action attribute values corresponding to the sequence of sub-actions;

input the first action attribute values and the time periods of the first action attribute values of each sub-action to a neural network model;

determine, using the neural network model, an action identifier for the sequence of sub-actions;

search, based on the action identifier via the one or more I/O paths, a content database for a plurality of mappings associated with respective video segments;

compare, using the neural network model, each mapping of the plurality of mappings to the first mapping;

determine, based on the comparing using the neural network model, a second mapping that matches the first mapping, wherein the second mapping is associated with a second video segment depicting a second character performing the action, and wherein the second mapping indicates second action attribute values and time periods of the second action attribute values corresponding to the second character performing the action;

generate, based on the first and the second mappings, a third mapping that indicates the second action attribute values and time periods of the second action attribute values, wherein the third mapping is associated with a video segment depicting the first character performing the action based on the second action attribute values and time periods of the second action attribute values;

retrieve a content structure comprising an object corresponding to the first character, wherein the content structure comprises virtual modelling data corresponding to the first character; and

modify the virtual modelling data based on the third mapping.

9. The system of claim 8 , wherein the control circuitry is further configured to generate, based on the modified virtual modelling data, a reconstructed first character performing the action based on the second action attribute values and time periods of the second action attribute values.

10. The system of claim 8 , wherein the control circuitry is further configured to:

generate a first fingerprint based on the first mapping; and

wherein the control circuitry is configured to compare the each mapping of the plurality of mappings to the first mapping by:

generating a respective fingerprint based on the each mapping; and

comparing the first fingerprint and the respective fingerprint.

11. The system of claim 8 , wherein the action identifier comprises an action keyword based on the sequence of sub-actions.

12. The system of claim 8 , wherein the content database comprises deconstructed video segments depicting expert actions.

13. The system of claim 8 , wherein the control circuitry is further configured to generate for display visual representations of the each mapping of the plurality of mappings.

14. The system of claim 13 , wherein the visual representations are displayed on a graphical interface.

Assignments (2)

Continuity (4)

References Cited (115)

Cited By (1)

US 12,718,564