Classification method

Patent metadata

Field	Value
Publication number	US-12080104-B2
Application number	US-202017637563-A
Country	US
Kind code	B2
Filing date	Sep 1, 2020
Priority date	Sep 12, 2019
Publication date	Sep 3, 2024
Grant date	Sep 3, 2024

Abstract

Person's conditions are classified according to his/her eye data. Person's conditions are classified according to his/her eye data using an imaging device, a feature extraction unit, and a classifier. The imaging device has a function of generating a group of images by continuous image capturing, and the group of images preferably includes an image of an eye area. The eye includes a black area and a white area. The method includes the steps in which the feature extraction unit extracts the eye area from the group of images, extracts a blinking amplitude, detects an image for determining start of eye blinking, stores an image for determining end of eye blinking as first data, and stores an image after a predetermined time elapsed from the first data as second data. The step in which the feature extraction unit extracts the white area from the first data and the second data is included. The classifier can use the white area as learning data.

First claim

The invention claimed is:

1. A classification method using an imaging device, a feature extraction unit, and a classifier,
wherein the classifier comprises a classification model,
wherein the imaging device has a function of generating a group of images by continuous image capturing,
wherein the group of images comprises an image of an eye area,
wherein the eye comprises a white area,
wherein the white area is an area in which an eyeball is covered with a white coating,
wherein the classification method comprises the steps in which the feature extraction unit extracts the eye area from the group of images, extracts a blinking amplitude from the group of images, detects an image for determining start of eye blinking from the group of images, stores an image for determining end of eye blinking from the group of images as first data, and stores an image after a predetermined time elapsed from the first data from the group of images as second data,
wherein the classification method comprises the step in which the feature extraction unit extracts area data of the white area from the first data and the second data,
wherein the classification method comprises the step in which the feature extraction unit supplies the area data of the white area to the classifier as learning data, and
wherein the classification method comprises the step in which the classifier generates the classification model using the learning data.

2. The classification method according to claim 1 ,
wherein the images of the eye areas included in the first data and the second data each include the white area,
wherein in the case where the detected area data of the white area comprises a first area and a second area independently, the classification method comprises the step in which the feature extraction unit outputs a ratio between the first area and the second area,
wherein in the case where the detected area data of the white area is detected as a third area, the classification method comprises the steps in which the feature extraction unit detects a circle area from the third area, obtains a center of the circle from the circle area, divides the third area into the first area and the second area using an x-coordinate of the center of the circle, and outputs the ratio between the first area and the second area,
wherein the classification method comprises the step in which the feature extraction unit calculates an oscillation width of a white part of the eye from the ratio between the first area and the second area, and
wherein the classifier uses the oscillation width of the white part of the eye as the learning data.

3. The classification method according to claim 1 , comprising:
the step in which the feature extraction unit supplies the oscillation width of the white part of the eye and the blinking amplitude as the learning data to the classifier; and
the step in which the classifier generates the classification model using the oscillation width of the white part of the eye and the blinking amplitude.

4. The classification method according to claim 1 , comprising:
the step of training the classifier using the learning data;
the step of supplying new first data and new second data to the feature extraction unit; and
the step in which the classifier classifies person's emotional or physical conditions using the classification model.

5. The classification method according to claim 1 , comprising:
the step of attaching a teacher label to the learning data, and
the step in which the classifier trains the classifier using the learning data to which the teacher label is attached.

Assignees

• Semiconductor Energy Lab

Inventors

• Oguni Teppei
• Fukutome Takahiro

Classifications

• G06T2207/30201

  Face · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/10016

  Video; Image sequence · CPC title

• G06T7/60

  Analysis of geometric attributes · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title