Granted Patent B2

US 12,265,947 · App. 17/811,273 · Granted Apr 1, 2025

Inventors: Scott Bokma (San Francisco, CA); Brian J. Lonsdorf (Moss Beach, CA); Sönke Rohde (San Francisco, CA); Alan Martin Ross (San Francisco, CA); Michael Sollami (Cambridge, MA); David Woodward (Mountain View, CA); Jessica Lundin (Bellevue, WA)

Assignee: Salesforce, Inc.

G06Q10/101G06T11/60

Abstract

Claims (66)

1. A method comprising:

receiving, by a computing device, an input file, wherein the input file includes a plurality of entries, wherein different entries are associated with a different geographic region characteristic;

selecting, by the computing device, a guide for entries of the plurality of entries by determining a respective geographic region characteristic stored with the entry, wherein a guide includes guidelines for formatting content in the respective entry in the plurality of entries according to guidelines of a geographic region;

inputting, by the computing device, an image from the entry into a model trained to generate image variations of the image by varying a characteristic of the image, wherein the image variations depict variations of the image and are not included in the input file;

determining, by the computing device, bounding boxes for placing text in the image and the image variations, wherein bounding boxes are determined for the image and the image variations based on respective content of the image and the image variations;

analyzing, by the computing device, the image and the image variations with respective bounding boxes to place text items from the entry in the image and the image variations, wherein the text items are formatted in the image and the image variations based on the guidelines of the geographic region in the guide for the entry;

generating, by the computing device, a plurality of design variants for the entry based on the image and the image variations that include the text items; and

determining, by the computing device, information inserted in the input file indicating image variations or text items that are reusable for another entry in the plurality of entries with a different geographic region characteristic, wherein the image variations or text items are reused to generate design variants for the another entry.

2. The method of claim 1 , wherein the input file associates an image and a text item for the entry.

3. The method of claim 1 , wherein a plurality of design variants for the entry is generated automatically upon receiving the input file.

4. The method of claim 1 , wherein selecting the guide comprises:

receiving a selection of the guide from a user for an entry or automatically determining the guide for the entry.

5. The method of claim 1 , wherein analyzing the image comprises:

analyzing an object in the image; and

generating the image variations of the image based on a position of the object in the image.

6. The method of claim 5 , wherein analyzing the image comprises:

automatically cropping the image in different ways based on the object.

7. The method of claim 6 , wherein analyzing the image comprises:

generating the image variations using a generative adversarial network that:

receives the image and a plurality of random numbers;

generates a number of candidate image variations; and

selects a portion of the number of candidate image variations as the image variations.

8. The method of claim 1 , wherein determining bounding boxes comprises:

inputting the image and image variants into a second model that detects objects in the image and image variants and generates bounding boxes for the image and the image variants, wherein a bounding box that covers an object is not determined as a bounding box for the image or image variants.

9. The method of claim 8 , wherein analyzing the image and the image variations to place text items comprises:

inserting the text item in the bounding box in the image and the image variations.

10. The method of claim 9 , wherein analyzing the image and the image variations to place text items comprises:

varying the format of the text items in the image and the image variations.

11. The method of claim 1 , wherein the design variants for different entries in the plurality of entries are formatted differently based on respective geographic regions in the guides for the entries.

12. The method of claim 1 , wherein:

the image and the image variations are generated in a first batch of processing, and

after generating the image and the image variations in the first batch of processing, placing text items in the image and the image variations in a second batch of processing.

13. The method of claim 1 , further comprising:

generating a plurality of design variants for other entries in the plurality of entries in parallel with the entry based on receiving the input file, wherein the other entries are associated with respective geographic regions in the guides.

14. The method of claim 1 , further comprising:

receiving feedback for a design variant in the plurality of design variants; and

adjusting logic that is used to create design variants based on the feedback.

15. The method of claim 1 , further comprising:

outputting the plurality of design variants; and

receiving a selection of a design variant for the entry to use in creating content for the entry.

16. The method of claim 1 , wherein:

entries in the plurality of entries are associated with different geographic regions; and

a respective guide for the entry is associated with a brand.

17. A non-transitory computer-readable storage medium having stored thereon computer executable instructions, which when executed by a computing device, cause the computing device to be configurable to cause:

receiving an input file, wherein the input file includes a plurality of entries, wherein different entries are associated with a different geographic region characteristic;

selecting a guide for entries of the plurality of entries by determining a respective geographic region characteristic stored with the entry, wherein a guide includes guidelines for formatting content in the respective entry in the plurality of entries according to guidelines of a geographic region;

inputting an image from the entry into a model trained to generate image variations of the image by varying a characteristic of the image, wherein the image variations depict variations of the image and are not included in the input file;

determining bounding boxes for placing text in the image and the image variations, wherein bounding boxes are determined for the image and the image variations based on respective content of the image and the image variations;

analyzing the image and the image variations with respective bounding boxes to place text items from the entry in the image and the image variations, wherein the text items are formatted in the image and the image variations based on the guidelines of the geographic region in the guide for the entry;

generating a plurality of design variants for the entry based on the image and the image variations that include the text items; and

determining information inserted in the input file indicating image variations or text items that are reusable for another entry in the plurality of entries with a different geographic region characteristic, wherein the image variations or text items are reused to generate design variants for the another entry.

18. The non-transitory computer-readable storage medium of claim 17 , wherein analyzing the image comprises:

analyzing an object in the image; and

generating the image variations of the image based on a position of the object in the image.

19. The non-transitory computer-readable storage medium of claim 17 , wherein analyzing the image and the image variations to place text items comprises:

determining a bounding box in the image and the image variations for a text item from the entry.

20. An apparatus comprising:

one or more computer processors; and

a computer-readable storage medium comprising instructions for controlling the one or more computer processors to be configurable to cause:

receiving an input file, wherein the input file includes a plurality of entries, wherein different entries are associated with a different geographic region characteristic;

selecting a guide for entries of the plurality of entries by determining a respective geographic region characteristic stored with the entry, wherein a guide includes guidelines for formatting content in the respective entry in the plurality of entries according to guidelines of a geographic region;

inputting an image from the entry into a model trained to generate image variations of the image by varying a characteristic of the image, wherein the image variations depict variations of the image and are not included in the input file;

determining bounding boxes for placing text in the image and the image variations, wherein bounding boxes are determined for the image and the image variations based on respective content of the image and the image variations;

analyzing the image and the image variations with respective bounding boxes to place text items from the entry in the image and the image variations, wherein the text items are formatted in the image and the image variations based on the guidelines of the geographic region in the guide for the entry;

generating a plurality of design variants for the entry based on the image and the image variations that include the text items; and

determining information inserted in the input file indicating image variations or text items that are reusable for another entry in the plurality of entries with a different geographic region characteristic, wherein the image variations or text items are reused to generate design variants for the another entry.

Assignments (1)

Continuity (1)

References Cited (67)