CLOTHING MATCHING SYSTEM AND METHOD

• US 20170083789A1
• Filed: 05/20/2016
• Published: 03/23/2017
• Est. Priority Date: 09/22/2015
• Status: Active Grant

• Current Assignee: Shantha Pathak, Swati Shah
• Original Assignee: Shantha Pathak, Swati Shah
• Inventors: Shah, Swati, Pathak, Shantha
• Granted Patent: US 9,811,762 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: G06V 10/774 Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06V 40/10 Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands

Abstract

A system and method for clothing matching is provided. The clothing matching system automatically processes the captured image of the user once the image is captured and identifies a face region, a torso region, and a bottom region of the user from the captured image. The clothing matching system further identifies (i) colour coordination and contrast of the captured image, (ii) a dress colour and a pattern of clothes of the user in the captured image, (iii) a body type of the user, and (iv) a skin tone of the user in the captured image. The clothing matching system communicates information related to the captured image to the server. The server receives and compares the information related to the captured image of the user with the training images/patterns stored in the ML model to compute a score for the captured image of the user for the dress combination.

Claims

1. A clothing matching system for computing a score associated with a dress combination for a user, wherein said clothing matching system comprises:

a memory that stores a database and a first set of modules, wherein said database comprises at least one of (i) annotations and a computed score for a captured image of said user, and (ii) a set of rules to compute said score for said captured image;

a processor that executes said first set of modules, wherein said first set of modules comprises;

an image recognition module, executed by said processor, analyses an image of said user when said user wearing clothes for evaluation, wherein said image recognition module automatically recognizes said captured image of said user once said image is captured through a first computing device, wherein said image recognition module comprises;

a colour coordination module, executed by said processor, detects a colour and a pattern type of said clothes that worn by said user in said captured image;

a fit analysis module, executed by said processor, detects whether said clothes worn by said user in said captured image suit a body type of said user and/or fit said user; and

a skin tone detection module, executed by said processor, detects a skin tone of said user in said captured image;

an image information communication module, executed by said processor, communicates characteristics information related to said captured image of said user to a server to compute a score associated with said dress combination for said captured image of said user, wherein said server comprises a machine-learning (ML) model that stores (i) trained images/patterns, (ii) a score provided by a plurality of human labellers for each of training images, (iii) annotations associated with each of said training images, (iv) a set of rules to compute a score for a captured image of a user, wherein said server (i) computes said score associated with said dress combination for said captured image of said user by comparing said characteristics information of said captured image with said annotated training images and/or patterns that are stored in said machine learning (ML) model, and (ii) communicates said computed score associated with said dress combination for said user to said clothing matching system; and

a score receiving module, executed by said processor, that (i) receives said computed score for said captured image of said user from said server, and (ii) displays said received score to said user through said first computing device.

2. The clothing matching system of claim 1, wherein said image recognition modules further comprises a body part demarcation module, executed by said processor, identifies and demarcates a face region, a torso region, a bottom region and/or a leg/foot region of said user in said captured image.

3. The clothing matching system of claim 1, wherein said server comprises:

a server memory that stores a server database and a second set of modules, wherein said server database comprises said machine-learning (ML) model that stores at least one of (i) trained images/patterns, (ii) a score provided by a plurality of human labellers for each of training images, (iii) annotations associated with each of said training images, (iv) a set of rules to compute a score for a captured image of a user, and (v) a computed score for dress combination for said captured image of said user;

a server processor that executes said second set of modules, wherein said second set of modules comprises;

an image information receiving module, executed by said server processor, that receives said characteristics information related to said captured image of said user from said clothing matching system to compute said score associated with said dress combination for said captured image of said user; and

a scoring module, executed by said server processor, that (i) computes said score for dress combination for said captured image of said user by comparing said characteristics information of said captured image with said annotated training images and/or patterns that are stored in said machine learning (ML) model, and (ii) communicates said computed score for said dress combination to said clothing matching system.

4. The clothing matching system of claim 1, wherein said image recognition module automatically processes said captured image of said user to detect said characteristics information that comprises at least one of (i) a colour coordination and contrast of said captured image, (ii) a dress colour and a pattern type of clothes, (iii) a body type of said user, (iv) whether said clothes worn by said user in said captured image suit a body type of said user and/or fit said user, and (v) a skin tone of said user.

5. The clothing matching system of claim 1, wherein said image recognition module identifies a gender and an age of said user from said captured image through an image-processing component.

6. The clothing matching system of claim 5, wherein said image information communication module communicates said gender and said age information of said user to said server to compute a score associated with said dress combination for said captured image of said user.

7. The clothing matching system of claim 1, wherein said score receiving module receives (i) said score for said captured image of said user, and (ii) a feedback/annotation for said captured image of said user from said server.

8. The clothing matching system of claim 1, wherein said clothing matching system automatically recognizes said captured image of said user using image processing techniques when said user captures an image through said first computing device.

9. The clothing matching system of claim 1, wherein said clothing matching system comprises an image-processing component to identify said characteristics information associated with said captured image of said user.

10. The clothing matching system of claim 1, wherein said system further comprises an image training system that provides a plurality of options to a plurality of human labellers to (a) demarcate (i) a face region, (ii) a torso region, and (iii) a bottom region of each of said training images, (b) provide a score for (i) said face region, (ii) said torso region, and (iii) said bottom region of each of said training images based on characteristics of demarcated regions that comprising at least one of a colour coordination, a dress colour, a pattern of clothes, a body type, (iii) provide a score for each of said training images based on said characteristics of said training images.

11. The clothing matching system of claim 10, wherein said image training system provides an option to said plurality of human labellers to annotate training images/patterns as at least one of, but not limited to (i) good or bad dressed people, (ii) dashing, (iii) gorgeous, (iv) dull, (v) passable, or (vi) not quite based on said characteristics of said captured image.

12. The clothing matching system of claim 3, wherein said image information receiving module receives information related to a gender type and an age of said user from said clothing matching system.

13. The clothing matching system of claim 3, wherein said scoring module (i) computes (a) a cumulative score for said entire captured image of said user, and (b) an individual scores for said face region, said torso region, and said bottom region of said captured image of said user, (ii) provide feedback/annotation on whether said dress combination of said user in said captured image as at least one of, but not limited to (a) good or bad, (b) dashing, (c) gorgeous, (d) dull, (e) passable, or (f) not quite, and (iii) communicate (i) said feedback/annotation, and (ii) said score for said captured image to said clothing matching system.

14. A computer implemented method for detecting characteristics information related to a captured image of a user to compute a score for dress combination using clothing matching system, said method comprising:

capturing an image of said user when said user wearing clothes for evaluation;

recognizing said image of said user when said image is captured using a first computing device;

identifying and demarcating a face region, a torso region, and a bottom region of said captured image;

detecting a colour and pattern type of clothes that said user wearing in said captured image;

detecting whether said clothes worn by said user in said captured image suit body type of said user;

detecting a skin tone of said user in said captured image;

communicating said characteristics information related to said captured image that comprises at least one of (i) a colour coordination and contrast of said clothes, (ii) a dress colour and pattern type of said clothes, (iii) a body type of said user, (iv) whether said clothes worn by said user in said captured image suit a body type of said user and/or fit said user, and (v) a skin tone of said user to a server, wherein said server comprises a machine-learning (ML) model that stores (i) trained images/patterns, (ii) a score provided by a plurality of human labellers for each of training images, (iii) annotations associated with each of said training images, (iv) a set of rules to compute a score for a captured image of a user, wherein said server (i) computes said score associated with said dress combination for said captured image of said user by comparing said characteristics information of said captured image with said annotated training images and/or patterns that are stored in said machine learning (ML) model, and (ii) communicates said computed score associated with said dress combination for said user to said clothing matching system; and

receiving a computed score for said captured image of said user from said server to display said computed score to said user through said first computing device.

15. The computer implemented method of claim 14, wherein said server performing steps of:

receiving said characteristics information associated with said captured image that comprising at least one of (i) said colour coordination and contrast of said clothes, (ii) said dress colour and pattern type of said clothes, (iii) said body type of said user, (iv) whether said clothes worn by said user in said captured image suit said body type of said user and/or fit said user, and (v) said skin tone of said user from said clothing matching system;

computing said score for dress combination for said captured image of said user by comparing (i) said colour coordination and contrast of said clothes, (ii) said dress colour and pattern type of said clothes, (iii) said body type of said user, (iv) whether said clothes worn by said user in said captured image suit said body type of said user and/or fit said user, and (v) said skin tone of said user in said captured image with trained images/patterns that are stored in a machine-learning (ML) model; and

communicating said computed score associated with said dress combination for said user to said clothing matching system.

16. The computer implemented method of claim 15, further comprising steps to provide a score for said training image using an image training system that performing steps of:

demarcating a face region, a torso region, and a bottom region of a plurality of training images;

detecting a colour and a pattern type of clothes that a person worn in said training image;

detecting whether clothes worn by said person in said training image suit a body type of said person;

detecting a skin tone of said person in said training image;

providing a score for dress combination for said training image; and

providing an annotation for dress combination on said training image.

17. The computer implemented method of claim 14, further comprising steps of:

detecting a gender, and an age of said user from said captured image; and

communicating said information related to said gender, and said age of said user to said server; and

receiving annotations for said captured image of said user from said server.

18. The computer implemented method of claim 15, further comprising steps of:

receiving said information related to a gender, and an age of said user from said clothing matching system;

computing said score for (i) a face region, a torso region, and a bottom region of said captured image, or (ii) said entire captured image of said user;

providing annotation for said captured image whether said dress combination of said user in said captured image is good or bad, dashing, gorgeous, dull, passable, or not quite; and

communicating said annotation for said captured image to said clothing matching system.

19. One or more non-transitory computer readable storage mediums storing one or more sequences of instructions, which when executed by one or more processors, causes computing a score associated with dress combination for a user by processing characteristics information related to a captured image, by performing the steps of:

receiving said characteristics information associated with said captured image that comprising at least one of (i) a colour coordination and contrast of clothes, (ii) a dress colour and pattern type of said clothes, (iii) a body type of said user, (iv) whether said clothes worn by said user in said captured image suit said body type of said user and/or fit said user, (v) a skin tone of said user, and (vi) an information related to a gender, and age of said user from a clothing matching system;

receiving a demarcated face region, a torso region, and a bottom region of said captured image from said clothing matching system;

computing said score for dress combination for said captured image of said user by comparing (i) said colour coordination and contrast of said clothes, (ii) said dress colour and pattern type of said clothes, (iii) said body type of said user, (iv) whether said clothes that worn by said user in said captured image suit said body type of said user and/or fit said user, and (v) said skin tone of said user in said captured image with trained images/patterns that are stored in a machine-learning (ML) model; and

computing said score for (i) a face region, a torso region, and a bottom region of said captured image, or (ii) said entire captured image of said user;

providing annotation for said dress combination of said user in said captured image; and

communicating (i) said computed score for said dress combination for said user, and (ii) annotation for said captured image of said user to said clothing matching system.

20. The one or more non-transitory computer readable storage mediums storing one or more sequences of instructions of claim 19, wherein said clothing matching system performing steps of:

capturing an image of said user when said user wearing clothes for evaluation;

recognizing said image of said user when said image is captured using a first computing device;

identifying and demarcating a face region, a torso region, and a bottom region of said captured image;

detecting a colour and pattern type of clothes that said user wearing in said captured image;

detecting whether said clothes worn by said user in said captured image suit body type of said user;

detecting a skin tone of said user in said captured image;

detecting a gender, and an age of said user from said captured image;

communicating said characteristics information related to said captured image that comprising at least one of (i) said colour coordination and contrast of said clothes, (ii) said dress colour and pattern type of said clothes, (iii) said body type of said user, (iv) whether said clothes worn by said user in said captured image suit a body type of said user and/or fit said user, (v) said skin tone of said user, and (vi) said information related to said gender, and said age of said user to a server; and

receiving (i) said computed score for said captured image of said user, and (ii) annotations for said captured image of said user from said server to display said computed score and annotations to said user through said first computing device.