# US 2021/0068404 A1 · Method for evaluating a health state of an anatomical element, related evaluation device and related evaluation system

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0068404 A1
- Document ID: US20210068404A1
- Status: Published application
- Kind: A1
- Application number: 17/009,973
- Filed: 2020-09-02
- Published: 2021-03-11
- Earliest claimed: 2019-09-09
- Assignees: Farm4Trade S.R.L.
- Applicants: Farm4Trade S.R.L.
- Inventors: Giuseppe MARRUCHELLA; Luca BERGAMINI; Andrea CAPOBIANCO DONDONA; Ercole DEL NEGRO; Francesco DI TONDO; Angelo PORRELLO; Simone CALDERARA
- Web page: https://patentgrep.com/patent/US20210068404A1

## CPC

- A22B 5/007: {Non-invasive scanning of carcasses, e.g. using image recognition, tomography, X-rays, ultrasound (analysing meat, e.g. search for foreign embedded objects G01N33/12; image analysis G06T7/00)}
- A22C 17/008: {for measuring quality, e.g. to determine further processing}
- G01N 33/12: Meat; Fish
- G06N 20/00: Machine learning
- G06T 7/0004: {Industrial image inspection}
- G06T 7/0014: {using an image reference approach}
- G06T 2207/20076: Probabilistic image processing
- G06T 2207/20081: Training; Learning
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30096: Tumor; Lesion
- G06T 2207/30128: Food products
- A22B: SLAUGHTERING
- A22C: PROCESSING MEAT, POULTRY, OR FISH (preserving A23B; obtaining protein compositions for foodstuffs A23J1/00; fish, meat or poultry preparations A23L; disintegrating, e.g. chopping meat, B02C18/00; preparation of proteins C07K1/00)
- G01N: INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES (measuring or testing apparatus or processes other than immunoassay, involving enzymes or microorganisms C12M, C12Q)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

A method for evaluating (50) the health state of an anatomical element of an animal in a slaughtering plant provided with an image acquisition device. The evaluation method comprises the steps of: verifying the presence of the anatomical element; acquiring the image of the anatomical element; processing (S4) the image of the anatomical element through Deep Learning techniques, generating a lesion image representing lesioned portions of the anatomical element, and a number of processed images, each representing a corresponding non-lesioned anatomical area of the animal; for each of the lesioned portions and for each of the non-lesioned anatomical areas, determining a corresponding quantity indicative of the probability that said lesioned portion corresponds to said non-lesioned anatomical area; determining a score indicative of the health state of the anatomical element, depending on the determined quantities.

## Claims

### Claim 1

A method for evaluating (50) the health state of an anatomical element (10) of an animal in a slaughtering plant (1), the slaughtering plant (1) being provided with an image acquisition device (18, 22) arranged so as to acquire an image of an observation area (20) of the slaughtering plant (1), the evaluation method (50) comprising the steps of:
verifying (S2), through the image acquisition device (18, 22), the presence of the anatomical element (10) in the observation area (20);
if the presence of the anatomical element (10) in the observation area (20) is verified, acquiring (S3), through the image acquisition device (18, 22), the image of the anatomical element (10);
processing (S4), by an image processing unit (34; 40) operatively coupled to the image acquisition device (18, 22), the image of the anatomical element (10) through deep learning techniques, generating a lesion image (P6), which represents lesioned portions of the anatomical element (10), and a number of processed images (P4, P5), each of which represents a corresponding non-lesioned anatomical area of the animal;
for each of the lesioned portions and for each of the non-lesioned anatomical areas, determining (S5) a corresponding quantity indicative of the probability that said lesioned portion corresponds to said non-lesioned anatomical area;
determining (S6) a score indicative of the health state of the anatomical element (10), depending on the determined quantities.

### Claim 2 (depends on 1)

The evaluation method according to claim 1, wherein the step of processing (S4) the image of the anatomical element (10) through the deep learning techniques comprises the steps of:
resizing (S4a) the image generating a resized image (Iresized) having at least one predefined size;
processing (S4a) the resized image (Iresized) through a first portion of a neural network, identifying a plurality of features of the resized image (Iresized); and
processing (S4b) the plurality of features of the resized image (Iresized) through a second portion of the neural network, generating at least the lesion image (P6) and the number of processed images (P4, P5), the second portion of the neural network having a transposed structure with respect to a structure of the first portion of the neural network.

### Claim 3 (depends on 2)

The evaluation method according to claim 2, further comprising the steps of:
processing training images using one or more data augmentation techniques generating a training data set; and
training the neural network based on the training data set through supervised training by means of a loss function.

### Claim 4 (depends on 1)

The evaluation method according to claim 1, wherein the step of determining (S5) each quantity comprises the steps of:
identifying, through a connected component technique, the number of lesioned portions in the lesion image (P6); and
comparing the lesioned portions and the non-lesioned anatomical areas to each other.

### Claim 5 (depends on 4)

The evaluation method according to claim 4, wherein the step of comparing the number of lesioned portions and the non-lesioned anatomical areas to each other comprises the step of calculating, for each lesioned portion in the lesion image (P6), a respective overlap area (Aintersection,k) between such lesioned portion and each non-lesioned anatomical area.

### Claim 6 (depends on 5)

The evaluation method according to claim 5, wherein the step of determining (S6) the score indicative of the health state of the anatomical element (10) comprises the steps of:
comparing, for each lesioned portion in the lesion image (P6), the respective overlap area (Aintersection,k) with an overlap threshold (Tintersection,k);
identifying as selected lesioned portions the lesioned portions whose overlap area (Aintersection,k) satisfies a predefined relation with the overlap threshold (Tintersection,k); and
determining the score based on the selected lesioned portions.

### Claim 7 (depends on 1)

The evaluation method according to claim 1, the anatomical element (10) being a carcass, the evaluation method being configured to detect pleurites in the carcass, and wherein the step of processing (S4) the image of the carcass through the deep learning techniques comprises the step of generating at least: an image of a first portion of hemithorax (P4), indicative of a first portion of hemithorax of the carcass; an image of a second portion of hemithorax (P5), indicative of a second portion of hemithorax of the carcass; and the lesion image (P6),
the processed image (P4, P5) being the image of the first portion of hemithorax (P4) and/or the image of the second portion of hemithorax (P5).

### Claim 8 (depends on 1)

The evaluation method according to claim 1, wherein the step of determining (S5) each quantity comprises the step of processing the lesion image (P6) and the number of processed images (P4, P5) through one of the following classifiers: support vector machine classifier, SVM; random forest classifier, RF; and decision tree classifier.

### Claim 9 (depends on 1)

The evaluation method according to claim 1, wherein the step of verifying (S2) the presence of the anatomical element (10) in the observation area (20) comprises the step of detecting the anatomical element (10) through at least one sensor (22) selected from: a photocell, or a sensor based on laser technology, or an acoustic sensor, or a snap-in sensor.

### Claim 10 (depends on 1)

The evaluation method according to claim 1, wherein the image acquisition device (18, 22) comprises a video camera, and the step of acquiring (S3) the image of the observation area (20) comprises the step of acquiring, through such video camera, a video of the observation area (20), one or more frames of this video forming the image of the observation area (20).

### Claim 11

An evaluation device (45′; 45″) to evaluate the health status of an anatomical element (10) of an animal in a slaughtering plant (1), the evaluation device (45′; 45″) being operatively couplable to an image acquisition device (18, 22) arranged so as to acquire an image of an observation area (20) of the slaughtering plant (1), the evaluation device (45′; 45″) comprising a control unit (36) operatively coupled to an image processing unit (34; 40),
wherein the control unit (36) is configured to:
verify, through the image acquisition device (18, 22), the presence of the anatomical element (10) in the observation area (20); and
if the presence of the anatomical element (10) in the observation area (20) is verified, acquire, through the image acquisition device (18, 22), the image of the anatomical element (10),
and wherein the image processing unit (34; 40) is configured to:
process the image of the anatomical element (10) through deep learning techniques, generating a lesion image (P6), which represents lesioned portions of the anatomical element (10), and a number of processed images (P4, P5), each of which represents a corresponding non-lesioned anatomical area of the animal;
for each of the lesioned portions and for each of said non-lesioned anatomical areas, determine a corresponding quantity indicative of the probability that said lesioned portion corresponds to said non-lesioned anatomical area; and
determine a score indicative of the health state of the anatomical element (10), depending on the determined quantities.

### Claim 12 (depends on 11)

The evaluation device according to claim 11, wherein the control unit (36) and the image processing unit (34; 40) are mutually coupled electromagnetically or through the Internet.

### Claim 13 (depends on 11)

An evaluation system (30′; 30″) for evaluating the health state of an anatomical element (10) of an animal in a slaughtering plant (1), the evaluation system (30′; 30″) comprising:
the evaluation device (45′; 45″), according to claim 11; and
the image acquisition device (18, 22), operatively coupled to the evaluation device (45′; 45″).
