Granted Patent B2

US 12,374,080 · App. 16/994,936 · Granted Jul 29, 2025

Inventors: Stephanie Sujin Choi (Seoul, KR); Hyeong In Choi (Seoul, KR)

G06V10/762A01K11/006G06F18/214G06F18/2411G06F18/2413G06V10/765G06V40/10

Abstract

Claims (54)

1. A computer-implemented method comprising instructions stored on a non-transitory computer readable storage medium and run on a computing device comprising memory and a hardware processor for obtaining an animal identification, the method comprising the steps of:

acquiring frontal view and side view of animal images based on first and second physical traits, wherein the traits are a qualitative description of a character of the animal, and the character is an abstract term referring to a characteristic of a specific body part of the animal, wherein the traits relates to at least two from the group consisting of head shape, ear shape, tail shape, relative eye and nose position, body profile contour, and nose pattern;

measuring a plurality of trait distances, wherein the trait distance is a measure of discrepancy between corresponding trait features of two animals, and the trait feature is a scalar or vector quantity that is a property of a trait of the animal;

computing a first ratio of the first trait which is muzzle length over a head length of the animal;

defining the first ratio as the head shape trait feature;

computing a first absolute value of a difference between the ratios of two said animals; and

defining the first absolute value of the difference as the head shape trait distance between two said animals,

computing a second ratio of the second trait of the animal;

defining the second ratio as the second trait feature;

computing a second absolute value of a difference between the ratios of two said animals;

defining the second absolute value of the difference in the second traits between the two animals;

identifying the two animals based on the absolute values of the first and second traits;

defining the body profile contour trait feature by a profile set;

computing the profile set by steps comprising:

resizing and rescaling the body profile contour to fit in a bounding box of a fixed size;

selecting pixels that lie within the resized and rescaled body profile contour;

defining the body profile contour trait distance between two said animals as normalized size of a symmetric difference between the profile sets of two said animals;

computing the normalized size of the symmetric difference of the profile sets of two said animals by steps comprising:

calculating the size of the symmetric difference between the profile sets of two said animals; and

dividing the size of the symmetric difference by the fixed size of the bounding box; and

outputting the identification of the two animals to a user.

2. The method of claim 1 , further comprising:

computing a combined distance function, wherein the combined distance function is a function of the plurality of trait distances.

3. The method of claim 2 , wherein the combined distance function is a weighted linear combination of the plurality of trait distances.

4. The method of claim 1 , further comprising:

capturing the image from a long distance for a surveillance-type identification.

5. The method of claim 1 , further comprising:

identifying the animal using a nose pattern biometric identification.

6. The method of claim 1 , wherein the animals are dogs, cats, cattle, or horses.

7. The method of claim 1 , further comprising:

computing an angle between a line connecting a base of the ear at a top of the head to a tip of the ear and a vertical line passing through the base of the ear;

defining the angle as the ear shape trait feature;

computing an absolute value of a difference between the angles of two said animals; and

defining the absolute value of the difference as the ear shape trait distance between two said animals.

8. The method of claim 1 , further comprising:

defining the tail shape trait feature by the vector quantity; and

determining components of the vector quantity by computing

a ratio of a tail length over a body length,

a ratio of an average tail thickness over the tail length, and

an angle between a horizontal line and a line connecting a start of the tail to an end of the tail.

9. The method of claim 8 , further comprising:

determining components of the vector quantity by computing

an absolute value of a difference between the ratios of the tail length over the body length of two said animals,

an absolute value of a difference between the ratios of the average tail thickness over the tail length of two said animals, and

an absolute value of a difference between the angles between the horizontal line and the line connecting the start of the tail to the end of the tail of two said animals.

10. The method of claim 1 , further comprising:

computing a bottom angle of a triangle formed by connecting centers of both eyes and a bottom tip of a philtrum;

defining the angle as the relative eye and nose position trait feature;

computing an absolute value of a difference between the bottom angles of two said animals; and

defining the absolute value of the difference as the relative eye and nose position trait distance between two said animals.

11. The method of claim 1 , further comprising:

computing a ratio of a leg length over a height; and

defining the ratio as the body profile contour trait feature.

12. The method of claim 2 , wherein the combined distance function is trained from data using a machine learning algorithm.

Continuity (2)

References Cited (21)