Granted Patent B2

US 12,694,705 · App. 18/375,262 · Granted Jul 28, 2026

Inventors: Michelle Salese (South Plainfield, NJ); Ronak Rughani (Edison, NJ); Katharine Podimatis (New York, NY)

Assignee: L'Oreal

G06V40/10G06V10/82A45D2044/007

Abstract

Claims (16)

1 . A computer-implemented method of training a neural network for hair curl detection, comprising:

collecting a set of digital coily hair pattern images, curly hair pattern images, kinky hair pattern images, or wavy hair pattern images from a data store;

applying one or more transformations to each digital hair image to create a modified set of digital hair images;

creating a first training set comprising the collected set of digital hair images, the modified set of digital hair images, and a set of digital non-hair curl pattern images;

training a neural network in a first stage using the first training set;

creating a second training set for a second stage of training comprising the first training set and digital non-hair curl images that are incorrectly detected as hair curl pattern images after the first stage of training; and

training the neural network in a second stage using the second training set.

2 . The computer-implemented method of claim 1 , wherein the collected set of digital hair images are labeled with users' answers to questions regarding users' characterization of curl patterns according to a curl pattern scale, and wherein the labeled digital hair images are used in training the neural network.

3 . A computer system comprising processing circuitry configured to:

collect a set of digital coily hair pattern images, curly hair pattern images, kinky hair pattern images, or wavy hair pattern images from a data store;

apply one or more transformations to each digital hair image to create a modified set of digital hair images;

create a first training set comprising the collected set of digital hair images, the modified set of digital hair images, and a set of digital non-hair curl pattern images;

train a neural network in a first stage using the first training set;

create a second training set for a second stage of training comprising the first training set and digital non-hair curl images that are incorrectly detected as hair curl pattern images after the first stage of training; and

train the neural network in a second stage using the second training set.

4 . The computer system of claim 3 further comprising processing circuitry configured to label the collected set of digital hair images with users' answers to questions regarding users' characterization of curl patterns according to a curl pattern scale, wherein the labeled digital hair images are used in training the neural network.

Assignments (1)

Priority Claims (1)

Continuity (2)

References Cited (28)