Egg Classification System

Abstract

To provide an egg classification system which is compact and can determine the state of the inside of an egg in a short time.SOLUTION: An egg classification system 10 for classifying an egg E comprises: a conveyance unit 11; an irradiation unit; a height change unit; an image data acquisition unit 21; a determination unit 22; and a decision unit 23. The conveyance unit 11 conveys a tray T that stores eggs E. The irradiation unit irradiates the eggs E stored in the tray T with light. The height change unit changes the height position of the irradiation unit according to the size of the egg E. The image data acquisition unit 21 acquires image data that expresses an image obtained by imaging the egg E in such a state that light is diffused inside with a color camera 12. The determination unit 22 determines the state of the inside of the egg E according to the image data acquired by the image data acquisition unit 21 with a learning model learned by using information expressing the state of the inside of the egg E and teacher data including the image data for the egg E. The decision unit 23 decides the classification destination of the egg E according to the determination result of the state of the inside of the egg E.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an egg sorting system for sorting eggs before chicks hatch. [Background technology]

[0002] A system for non-destructively inspecting and classifying the internal condition of eggs is known, such as that described in Patent Document 1 below. This system rotates eggs placed on an inspection foil and photographs them from four directions with a color camera. From the images captured by the color camera, information on the presence, thickness, and distribution of blood vessels, the color of the egg's interior, and the density near the air cell boundary are measured, and by comparing these with images of good eggs, the viability and developmental state of fertilized eggs is determined. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2004-101204 Summary of the Invention [Problem to be solved by the invention]

[0004] In the above-mentioned method, it takes a long time to inspect each egg, and the inspection equipment is large.

[0005] An object of the present invention is to provide an egg sorting system that is compact and capable of determining the internal condition of an egg in a short period of time. [Means for solving the problem]

[0006] The egg sorting system of the present invention is an egg sorting system for sorting eggs before they hatch into chicks, and comprises a transport unit, an illumination unit, a height change unit, an image data acquisition unit, a judgment unit, and a determination unit. The transport unit transports trays containing eggs. The illumination unit irradiates light onto the eggs contained in the trays. The height change unit changes the height position of the illumination unit depending on the size of the eggs. The image data acquisition unit acquires image data representing images of eggs with light diffused inside them taken by a color camera. The judgment unit , painting The image data acquisition unit determines the internal state of the egg based on the image data acquired. The determination unit determines the classification of the egg based on the determination result of the internal state of the egg. [Effects of the Invention]

[0007] According to the present invention, it is possible to provide an egg sorting system that is compact and capable of determining the internal condition of an egg in a short period of time.

[0008] The above and other objects, features, aspects and advantages of the present invention will become apparent from the following detailed description of the invention taken in conjunction with the accompanying drawings. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a side view showing a schematic diagram of an egg sorting system according to an embodiment of the present invention; [Figure 2] FIG. 2 is an enlarged side view showing a main part of the egg sorting system according to the embodiment. [Figure 3] FIG. 1 is a schematic diagram showing a neural network. [Figure 4] FIG. 2 is a functional block diagram of the classification device according to the embodiment. [Figure 5] 10 is a flowchart showing the procedure of a setting process. [Figure 6] 10 is a flowchart showing a processing procedure for classifying eggs. DETAILED DESCRIPTION OF THE INVENTION

[0010] An egg sorting system 10 according to this embodiment will be described below with reference to FIGS.

[0011] The egg sorting system 10 according to this embodiment uses images to sort eggs E before they hatch into chicks. The eggs E are placed on a tray T called a setter tray T, which is used in an incubator. The tray T has a plurality of storage seats, for example, six rows horizontally and six rows vertically. The eggs E are placed in each storage seat on the tray T with their long axes facing vertically. The eggs E are stored in the tray T, with the upper half of the eggs E visible. The eggs E are photographed after being removed from the incubator. For example, the photograph is taken on the 10th day of incubation. The number of days of incubation is the number of days that have passed since the egg-setting date, which is the day the incubation process begins.

[0012] The egg sorting system 10 comprises an egg photographing device 1 and a sorting device 2. The egg photographing device 1 is mainly composed of a transport unit 11, an irradiation unit (not shown), and a height change unit (not shown). The egg photographing device 1 further comprises a color camera 12, an arm unit (not shown), a cap 13, a cover 14, and a control unit (not shown). The sorting device 2 is mainly composed of an image data acquisition unit 21, a judgment unit 22, and a decision unit 23. The sorting device 2 further comprises an information acquisition unit 24 and an adjustment unit 25.

[0013] The conveying unit 11 conveys the trays T containing the eggs E. The conveying unit 11 is, for example, a belt conveyor that supports the trays T from below.

[0014] The irradiation unit irradiates light onto the eggs E contained in the tray T. The irradiation unit is, for example, an LED. The irradiation unit irradiates the upper end of each egg E. The eggs E glow brightly as the light is diffused inside. The irradiation units are arranged to correspond to the container seats of the tray T, and for example, six of them are lined up in a row across the conveyance direction.

[0015] The height changing unit changes the height position of the irradiation unit according to the size of the egg E. The height changing unit may be a unit that moves the illumination unit in the vertical direction using a separate drive source, or may be a unit that moves the illumination unit in the vertical direction by deforming at least one of the cap and the arm unit without using a separate drive source.

[0016] The color camera 12 photographs the eggs E from above. The color camera 12 is fixed above the transport unit 11 and photographs multiple eggs E simultaneously. The color camera 12 may photograph one egg E or multiple eggs E may be used. The color camera 12 may photograph one egg E multiple times.

[0017] The arm unit has an irradiation unit attached to its tip. The arm unit is deformable. Specifically, the base end of the arm unit is fixed, and the position of the tip moves up and down as the arm unit deforms. In this embodiment, while the transport unit 11 is operating, a force is applied to the tip of the arm unit in a direction that moves the tip of the cap 13 away from the upper end of the egg E, and while the transport unit 11 is stopped, a force is applied to the tip of the arm unit in a direction that presses the tip of the cap 13 against the upper end of the egg E.

[0018] The cap 13 is attached around the irradiation unit. Specifically, the base end of the cap 13 is fixed to the arm unit, and the tip end is in contact with the upper end of the egg E. The cap 13 is flexible and light-blocking. It is preferable that the lower end of the cap 13 is set to a size that does not completely cover the portion of the egg E that corresponds to the air chamber.

[0019] The cover 14 prevents external light from entering the eggs E to be photographed. Specifically, the cover 14 covers at least the tray T, the eggs E, and the irradiation unit.

[0020] The control unit controls the transport unit 11 to stop transport of the tray T before photographing the eggs E. The control unit controls the irradiating units so that adjacent eggs E do not light up at the same time. Specifically, the control unit turns off the second, fourth, and sixth irradiating units of the six irradiating units when the first, third, and fifth irradiating units are irradiated, and on the other hand, turns off the first, third, and fifth irradiating units when the second, fourth, and sixth irradiating units are irradiated. The control unit may also control the irradiating units to change the light emission intensity depending on the number of days of incubation.

[0021] The image data acquisition unit 21 acquires image data representing an image of the egg E in a state where light is diffused inside the egg E, photographed by the color camera 12. The image data acquisition unit 21 is connected to the determination unit 22.

[0022] The determination unit 22 is configured using a computer. The internal state of the egg E is determined according to the image data acquired by the image data acquisition unit 21, using a learning model trained using teacher data including information representing the internal state of the egg E and image data about the egg E. The determination unit 22 is equipped with a learning model used to determine the internal state of the egg E according to image data representing a photographed image of the egg E. The learning model uses, for example, a neural network.

[0023] FIG. 3 is a conceptual diagram showing an example of the functional configuration of a learning model. The learning model uses a neural network including an input layer 31, an intermediate layer 32, and an output device, each having a plurality of nodes. The input layer 31 has a plurality of nodes to which image data is input. For example, the pixel value of each pixel included in the image is input to one of the nodes. The output layer 33 has a plurality of nodes that receive data calculated from the nodes in the intermediate layer 32 and output the state of the egg E. For example, one node outputs the probability that the egg is unfertilized as a score, another node outputs the probability that the blood vessels are underdeveloped as a score, and yet another node outputs the probability that the air cell is improperly positioned as a score. The intermediate layer 32 may have one layer or multiple layers. The learning model may use a convolutional neural network (CNN) or a recurrent neural network (RNN) as the neural network.

[0024] The learning model is trained in advance using training data. Image data is acquired for multiple eggs E whose states are known, including unfertilized eggs, and training data is created. The training data includes information representing the internal state of each egg E and image data representing photographs of each egg E. Image data for each egg E included in the training data is input to each node in the input layer 31, and training is performed to adjust the calculation parameters of each node so that a score matching the state of each egg E indicated by the information included in the training data is output to each node in the output layer 33.

[0025] The determination unit 23 determines the classification destination of the egg E according to the determination result of the internal state of the egg E. Specifically, the image obtained when photographing the egg E differs depending on the state of development of the embryo. Therefore, it is possible to determine the internal state of the egg E according to the image data.

[0026] The information acquisition unit 24 acquires at least one of information related to eggshell color, information related to the breed of the chicken that will hatch, and information related to the number of days since the start of incubation. The egg sorting system 10 is equipped with an input device 15 into which this information is input. The input device 15 is operated by a user to input information. The information acquisition unit 24 may also automatically acquire this information from an incubator or other device configured to be able to communicate with the sorting device 2.

[0027] Eggshell color, chicken breed, and number of days since incubation began are related to the brightness of the egg E when exposed to light. Chicken breed is related to the color of the embryo inside the egg, and number of days since incubation began is related to the size of the embryo inside the egg. Therefore, information on eggshell color, chicken breed, and number of days since incubation began are factors that affect the inferred results when classifying egg E.

[0028] The adjustment unit 25 selects a learning model to be used by the determination unit 22 from among a plurality of learning models according to the acquired information, or adjusts the criteria for determining the classification of the egg E according to the determination result by the decision unit 23. For example, the image of the egg E changes slightly depending on the number of days that have passed since the start of incubation. The adjustment unit 25 receives information output from the input device 15 and adjusts the criteria for determining the classification of the egg E according to the received information.

[0029] The egg sorting system 10 also includes a rejection unit that rejects eggs E that will not be returned to the incubator from among the eggs E that have been sorted using the egg photography device 1 and the sorting device 2. The rejection unit has a mechanism for removing eggs E transported by the transport unit 11 from the tray T. Techniques well known in the art can be used to selectively transfer non-viable eggs E from the tray T.

[0030] Next, the procedure of the setting process performed by the classification device 2 will be described with reference to FIG.

[0031] The information acquisition unit 24 acquires at least one of information on the eggshell color, information on the breed of the chicken that will hatch, and information on the number of days that have passed since the start of incubation (steps S21, S22, S23). The information is set by the input device 15.

[0032] The criteria for determining the classification of eggs E need to be adjusted depending on the quality of the eggs E to be classified. By adjusting the criteria for determining the classification of eggs E, it becomes possible to appropriately classify eggs E. Alternatively, by adjusting the learning model used to determine the state of eggs E, it becomes possible to appropriately classify eggs E.

[0033] The adjustment unit 25 selects a learning model to be used in the next process (step S24). The learning model includes a plurality of learning models with different parameters, and the adjustment unit 25 selects a learning model to be used in the process from the plurality of learning models. The adjustment unit 25 also adjusts the criteria for classifying the eggs E according to the quality of the eggs E to be shipped (step S25). For example, the adjustment unit 25 changes the criteria for vascular underdevelopment used to determine the classification destination of the eggs E. Note that only one of step S24 and step S25 may be executed.

[0034] The setting process described above is performed for each lot, for example.

[0035] When the eggs E to be sorted are transported by the transport unit 11 to a predetermined photographing position, the transport unit 11 stops temporarily. When the height changing unit positions the illumination unit at the top end of the eggs, the control unit turns on the illumination unit and the eggs E are photographed by the color camera 12. When the photographing is complete, the control unit operates the transport unit 11, and when the eggs E in the next row are transported to the photographing position, the transport unit 11 stops again. When the illumination unit is positioned by the height changing unit, the illumination unit turns on and the color camera 12 photographs the eggs. This process is performed for all the eggs on the tray T.

[0036] Next, the procedure of the process performed by the determination unit 22 of the classification device 2 will be described with reference to FIG.

[0037] The determination unit 22 acquires image data of one egg E whose position has been identified from the image data acquisition unit 21 (step S1). Next, a setting process is performed to set the egg E for classification (S2). In step S2, the setting process described in steps S21 to S25 is performed.

[0038] Next, the determination unit 22 performs a determination process using a learning model based on the image data to determine the internal state of the egg E (step S3). The learning model has been trained in advance to determine the state of growth of the embryo inside the egg E from the image data, using training data including image data showing a photographed image of the egg E and information representing the internal state of the egg E.

[0039] The determination unit 22 determines the internal state of the egg E according to the output of the learning model. For example, the determination unit 22 determines that the egg is not an unfertilized egg when the probability that the egg is an unfertilized egg is less than a predetermined threshold, and determines that the egg is an unfertilized egg when the probability that the egg is an unfertilized egg is equal to or greater than the predetermined threshold.

[0040] The determination unit 22 outputs the determination result of the internal state of the egg E (step S4). The decision unit 23 determines the classification destination of the egg E according to the determination result of the internal state of the egg E (step S5). For example, the eggs E are classified into a classification destination for eggs E that will continue to be incubated and a classification destination for eggs E, including unfertilized eggs, for which incubation is to be discontinued. The determination unit 23 identifies the position on the tray T of the eggs E for which incubation is to be discontinued, and the exclusion unit excludes the eggs E for which incubation is to be discontinued from the tray T.

[0041] As described above, the egg sorting system 10 according to this embodiment sorts eggs E before they hatch, and includes the transport unit 11, the irradiation unit, the height change unit, the image data acquisition unit 21, the determination unit 22, and the decision unit 23. The transport unit 11 transports the trays T containing the eggs E. The irradiation unit irradiates the eggs E contained in the trays T with light. The height change unit changes the height position of the irradiation unit depending on the size of the eggs E. The image data acquisition unit 21 acquires image data representing images of the eggs E with light diffused inside, captured by the color camera 12. The determination unit 22 determines the internal state of the eggs E based on the image data acquired by the image data acquisition unit 21, using a learning model trained using training data including information representing the internal state of the eggs E and image data about the eggs E. The determination unit 23 determines the classification of the eggs E based on the determination result of the internal state of the eggs E.

[0042] The system is equipped with an information acquisition unit (24) that acquires at least one of information related to eggshell color, information related to the breed of the hatching chicken, and information related to the number of days elapsed since the start of incubation, and an adjustment unit (25) that selects a learning model to be used by the judgment unit (22) from among a plurality of learning models according to the acquired information, or adjusts the criteria for determining the classification of the egg (E) by the decision unit (23) according to the judgment result.

[0043] The device has an arm portion to the tip of which the irradiation portion is attached, and the height changing portion moves or deforms the arm portion.

[0044] The present invention is not limited to the above-described embodiment.

[0045] The egg sorting system 10 may be configured to determine the internal state of the eggs E outside the egg photographing device 1. For example, a sorting device 2 including a trained learning model may be provided outside the egg photographing device 1. The sorting device 2 is configured using a computer. The sorting device 2 may be configured using multiple computers, or may be realized using the cloud.

[0046] The height change unit may utilize the transport of the tray T. More specifically, this height change unit is provided with a roller at the tip of the arm unit that comes into contact with the upper end of the eggs E in the tray T. The roller is rotatably attached to the tip of the arm unit, and rides up over the eggs E as the tray T is transported. The roller is positioned so that its outer circumferential surface is aligned with the transport direction of the transport unit 11 and in a location that comes into contact with the upper end of the eggs E. The arm is held so that it can move up and down. A lighting unit is provided near the roller, and the height of the lighting unit can be changed via this roller depending on the size of the eggs E.

[0047] The irradiation unit is not limited to being attached to the tip of the arm unit, and the arm unit may not be elastically deformable.

[0048] The egg sorting system 10 of the present invention may be used on days other than the 10th day of incubation. For example, it is preferable to use it on days 9, 11, 18, and 19 of incubation.

[0049] The embodiments disclosed herein are examples and are not intended to be limiting. The present invention is defined not by the scope of the above description but by the scope of the claims, and is intended to include all modifications within the meaning and scope equivalent to the scope of the claims. [Industrial Applicability]

[0050] The present invention can be used in an egg sorting system. [Explanation of symbols]

[0051] 10...Egg Classification System 11...Transportation section 12...Color camera 21...Image data acquisition unit 22…Judgment section 23...Decision-making department 24…Information acquisition department 25...Adjustment section E…Egg T...tray

Claims

1. In the egg classification system, which classifies eggs before chicks hatch, a conveying unit that conveys trays containing eggs; an irradiation unit that irradiates light onto the eggs contained in the tray; an arm portion to which the irradiation portion is attached at a tip thereof; a height changing unit that changes the height position of the irradiation unit according to the size of the eggs by causing the tip of the arm unit to ride on top of the eggs as the tray is transported; and an image data acquisition unit that acquires image data representing an image of an egg with light diffused therein taken by a color camera; a determination unit that determines the internal state of the egg based on the image data acquired by the image data acquisition unit; a determination unit that determines the classification of the egg according to the determination result of the internal state of the egg; An egg sorting system comprising:

2. the determination unit determines the internal state of the egg using a learning model trained using teacher data including information representing the internal state of the egg and image data about the egg; Furthermore, an information acquisition unit that acquires at least one of information regarding the eggshell color, information regarding the breed of the chicken that will hatch, and information regarding the number of days that have passed since the start of incubation; 2. The egg sorting system according to claim 1, further comprising: an adjustment unit that selects a learning model to be used by the determination unit from among a plurality of learning models according to the acquired information; or an adjustment unit that adjusts a criterion for determining an egg classification according to the determination result by the determination unit.

3. An egg sorting system as described in claim 1 or 2, wherein the height changing unit moves or deforms the arm unit.

Images