Systems and methods for generating a visualization of workspace employees

Patent metadata

Field	Value
Publication number	US-12548217-B2
Application number	US-202318479534-A
Country	US
Kind code	B2
Filing date	Oct 2, 2023
Priority date	Oct 4, 2022
Publication date	Feb 10, 2026
Grant date	Feb 10, 2026

Abstract

Example implementations include a method, apparatus and computer-readable medium for generating a visual model of a uniform, comprising detecting a plurality of persons in one or more images. The implementations further include generating a sub-image depicting the respective person from the one or more images for each respective person of the plurality of persons. Additionally, the implementations further include identifying one or more articles of clothing worn by the respective person, executing at least one transformation that adjusts pixels of the sub-image such that the one or more articles of clothing depicted in the sub-image fit in a common template, calculating variance values between a plurality of transformed sub-images comprising the sub-image transformed by the at least one transformation and other sub-images generated for other persons in the plurality of persons and transformed by the at least one transformation, and generating a virtual model based on the variance values.

First claim

What is claimed is:

1 . An apparatus for generating a visual model of a uniform, comprising:
a memory; and a processor coupled with the memory and configured to:
detect a plurality of persons in one or more images;
generate, for each respective person of the plurality of persons, a sub-image depicting the respective person from the one or more images;
identify one or more articles of clothing worn by the respective person;
execute at least one transformation that adjusts pixels of the sub-image such that the one or more articles of clothing depicted in the sub-image fit in a common template;
calculate variance values between a plurality of transformed sub-images comprising the sub-image transformed by the at least one transformation and other sub-images generated for other persons in the plurality of persons and transformed by the at least one transformation by:
overlaying the sub-image with each of the other sub-images;
calculating a difference for each overlaid pixel value between the sub-image and each of the other sub-images; and
determining an average difference for each overlaid pixel value based on the difference between the sub-image and each of the other sub-images, wherein the variance values are a plurality of average differences; and

generate a virtual model depicting matching articles of clothing worn by the plurality of persons based on the variance values.

2 . The apparatus of claim 1 , wherein the processor is further configured to:
omit from the virtual model clothing in a region of a given sub-image with a respective variance value greater than a threshold variance value; and include in the virtual model the clothing in the region of the given sub-image with the respective variance value not greater than the threshold variance value.

3 . The apparatus of claim 1 , wherein the average difference is a weighted average difference.

4 . The apparatus of claim 1 , wherein the at least one transformation is one or more of: resizing, rescaling, relocating, upsampling, downsampling, or masking.

5 . The apparatus of claim 1 , wherein to execute the at least one transformation comprises to:
detect a graphic on the one or more articles of clothing; identify the graphic in a database of graphics; and adjust the pixels of the sub-image such that the one or more articles of clothing depicted in the sub-image fit in the common template and proportions of the graphic are not distorted.

6 . The apparatus of claim 1 , wherein the common template is a template boundary of a given shape and size, and wherein the pixels of the sub-image are adjusted such that a sub-image boundary of the one or more articles of clothing in the sub-image matches the template boundary of the common template.

7 . The apparatus of claim 1 , wherein to generate the sub-image depicting the respective person comprises to:
crop an image of the one or more images to an area that maximizes a depiction of the respective person and minimizes a depiction of objects that are not the respective person in the image, wherein respective pixels associated with the area are saved in the sub-image.

8 . The apparatus of claim 1 , wherein the processor is further configured to:
receive a text-based classification describing the virtual model; and input the text-based classification in a generative machine learning model that generates a visualization of the uniform corresponding to the virtual model.

9 . The apparatus of claim 1 , wherein the plurality of persons have a common uniform.

10 . A method for generating a visual model of a uniform, comprising:
detecting a plurality of persons in one or more images; generating, for each respective person of the plurality of persons, a sub-image depicting the respective person from the one or more images; identifying one or more articles of clothing worn by the respective person; executing at least one transformation that adjusts pixels of the sub-image such that the one or more articles of clothing depicted in the sub-image fit in a common template; calculating variance values between a plurality of transformed sub-images comprising the sub-image transformed by the at least one transformation and other sub-images generated for other persons in the plurality of persons and transformed by the at least one transformation by:
overlaying the sub-image with each of the other sub-images;
calculating a difference for each overlaid pixel value between the sub-image and each of the other sub-images; and
determining an average difference for each overlaid pixel value based on the difference between the sub-image and each of the other sub-images, wherein the variance values are a plurality of average differences; and
generating a virtual model depicting matching articles of clothing worn by the plurality of persons based on the variance values.

11 . The method of claim 10 , further comprising:
omitting from the virtual model clothing in a region of a given sub-image with a respective variance value greater than a threshold variance value; and including in the virtual model the clothing in the region of the given sub-image with the respective variance value not greater than the threshold variance value.

12 . The method of claim 10 , wherein the average difference is a weighted average difference.

13 . The method of claim 10 , wherein the at least one transformation is one or more of: resizing, rescaling, relocating, upsampling, downsampling, or masking.

14 . The method of claim 10 , wherein executing the at least one transformation comprises:
detecting a graphic on the one or more articles of clothing; identifying the graphic in a database of graphics; and adjusting the pixels of the sub-image such that the one or more articles of clothing depicted in the sub-image fit in the common template and proportions of the graphic are not distorted.

15 . The method of claim 10 , wherein the common template is a template boundary of a given shape and size, and wherein the pixels of the sub-image are adjusted such that a sub-image boundary of the one or more articles of clothing in the sub-image matches the template boundary of the common template.

16 . The method of claim 10 , wherein generating the sub-image depicting the respective person comprises:
cropping an image of the one or more images to an area that maximizes a depiction of the respective person and minimizes a depiction of objects that are not the respective person in the image, wherein respective pixels associated with the area are saved in the sub-image.

17 . The method of claim 10 , further comprising:
receiving a text-based classification describing the virtual model; and inputting the text-based classification in a generative machine learning model that generates a visualization of the uniform corresponding to the virtual model.

18 . The method of claim 10 , wherein the plurality of persons have a common uniform.

Assignees

• Sensormatic Electronics Llc

Inventors

• Root Cory Olog
• Wilson Matthew

Classifications

• G06T3/40

  Scaling of whole images or parts thereof, e.g. expanding or contracting · CPC title

• G06T2210/16

  Cloth · CPC title

• G06T7/337

  involving reference images or patches · CPC title

• G06T2210/22

  Cropping · CPC title

• G06V40/10

  Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands · CPC title