Cattle ischium estimation device, cattle evaluation system, and cattle ischium estimation program

Abstract

To accurately extract the ischium (ischium) portion of a cow. [Solution] The cow ischium estimation device comprises: an acquisition unit that acquires point cloud data of a cow; a first identification unit that approaches the point cloud data with a first plane along a first direction from the rear and above toward the cow and identifies a first point that the first plane first contacts; a second identification unit that excludes a three-dimensional region having a predetermined dimension in the width direction (second direction) of the cow and based on the first point, and identifies a first partial point cloud data that is not included in the excluded region and is located on one side of the excluded region in the second direction; and an extraction unit that approaches the first partial point cloud data with a second plane intersecting the third direction along a third direction which is from the rear and above toward the cow and identifies a second point that the second plane first contacts, and extracts a portion of the first partial point cloud data that is located within a predetermined range based on the second point as point cloud data for one ischium of the cow.

Description

[Technical Field]

[0001] The present invention relates to a device for estimating the ischial bone of a cow, a cow evaluation system, and a program for estimating the ischial bone of a cow. [Background technology]

[0002] To evaluate the energy balance in cattle management, the state of body fat attachment (muscle condition) of cattle is assessed. One numerical index indicating the state of body fat attachment is the Body Condition Score (BCS). Normally, the Body Condition Score is determined by veterinarians or dairy experts using methods such as the UV method, but there can be variability in the judgment between evaluators depending on the barn environment and the herd being evaluated, and even variability can occur among the same evaluator. On the other hand, a method is known in which a three-dimensional surface representation of a part of an animal is formed from a three-dimensional image of the animal, the surface of the three-dimensional surface representation is statistically analyzed, and a score for the animal's body condition is determined based on the analysis results (e.g., Patent Document 1). In addition, a method is known in which the user selects an arbitrary geodesic from three-dimensional coordinates of a dairy cow viewed from above, and calculates the hip angle width to estimate the weight of the dairy cow (e.g., Patent Document 2). [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Special Publication No. 2012-510278 [Patent Document 2] Japanese Patent Publication No. 2016-59300 [Overview of the project] [Problems that the invention aims to solve]

[0004] It is believed that leanness or weight gain in cattle affects the shape of localized parts of the animal. Therefore, if these parts can be accurately identified from three-dimensional images of the cattle, it is thought that a body condition score can be determined based on the three-dimensional shape of those parts. However, Patent Document 1 does not disclose a method for identifying localized parts that should be focused on when determining the body condition score. Furthermore, the method described in Patent Document 2 is cumbersome because it requires the user to select geodesics.

[0005] The present invention has been made in view of the above problems, and aims to provide a cattle ischium estimation device and cattle evaluation system, as well as a cattle ischium estimation program, that can accurately estimate the ischium of a cattle. [Means for solving the problem]

[0006] In the first embodiment, the cow ischial bone estimation device includes an acquisition unit that acquires point cloud data representing the three-dimensional shape of a cow in a standing position in three-dimensional space, a first identification unit that approaches the point cloud data along a first direction which is the direction toward the cow from behind and above in the three-dimensional space, and identifies a first point which the first plane first contacts, and an exclusion region which is a three-dimensional area having a predetermined dimension in a second direction which is the width direction of the cow and extending along the first direction with respect to the first point, and the exclusion The system includes: a second identification unit that identifies first partial point cloud data located on one side of the excluded region in the second direction from among the point cloud data not included in the region; and an extraction unit that, in the three-dimensional space, moves a second plane intersecting the third direction towards the cow along the third direction, which is the direction from behind and above the cow toward the cow, toward the first partial point cloud data, identifies a second point that the second plane first contacts, and extracts a portion of the first partial point cloud data located within a predetermined range based on the second point as point cloud data of one ischium of the cow.

[0007] In a second embodiment, the cattle evaluation system comprises a cattle ischial bone estimation device and an evaluation device that evaluates the meatiness of the cattle from index values ​​obtained from the three-dimensional shape shown by the point cloud data of the cattle ischial bone extracted by the cattle ischial bone estimation device.

[0008] In the third embodiment, the cow ischial bone estimation program acquires point cloud data representing the three-dimensional shape of a cow in a standing position in three-dimensional space, and in the three-dimensional space, it approaches the point cloud data with a first plane that intersects with a first direction, which is the direction toward the cow from behind and above, and identifies a first point that the first plane first contacts, and in the three-dimensional space, it excludes a three-dimensional region that has a predetermined dimension in the second direction, which is the width direction of the cow, and extends along the first direction with respect to the first point, and excludes the parts not included in the excluded region. This program causes a computer to perform the following steps: identify a first partial point cloud data set located on one side of the excluded region in the second direction from the point cloud data set; move a second plane that intersects the third direction in the three-dimensional space, along the third direction which is the direction from behind and above the cow toward the cow, approach the first partial point cloud data set; identify a second point that the second plane first contacts; and extract a portion of the first partial point cloud data set located within a predetermined range based on the second point as point cloud data for one ischium of the cow. [Effects of the Invention]

[0009] It can accurately estimate the location of the ischium in a cow. [Brief explanation of the drawing]

[0010] [Figure 1] Figure 1 is a schematic diagram showing the configuration of the cattle evaluation system according to the first embodiment. [Figure 2] Figure 2(a) shows an example of the hardware configuration of the device for estimating the area of ​​interest, and Figure 2(b) shows an example of the hardware configuration of the evaluation device. [Figure 3] Figure 3 is a functional block diagram of the device for estimating the area of ​​interest and the device for evaluation. [Figure 4] FIG. 4 is a flowchart showing the processing of the target site estimation device in the first embodiment. [Figure 5] FIG. 5(a) is a view showing the three-dimensional point cloud data of a cow as viewed from the side of the cow, and FIG. 5(b) is a view showing a portion corresponding to the three-dimensional point cloud data of FIG. 5(a) in the three-dimensional image of the cow with a white line frame. [Figure 6] FIG. 6(a) is a view showing the three-dimensional point cloud data of a cow as viewed from the rear of the cow. FIG. 6(b) is a view showing a portion corresponding to the three-dimensional point cloud data of FIG. 6(a) in the three-dimensional image of the cow with a white line frame. [Figure 7] FIG. 7 is a diagram for explaining the processing of step S16 in FIG. 4. [Figure 8] FIGS. 8(a) and 8(b) are diagrams for explaining the processing of step S18 in FIG. 4. [Figure 9] FIGS. 9(a) and 9(b) are diagrams for explaining a method of extracting the point cloud data of the right os coxae of a cow. [Figure 10] FIG. 10(a) is a view showing the state of the second point as viewed from the rear of the cow, and FIG. 10(b) is a view showing the state of the third point as viewed from the rear of the cow. [Figure 11] FIGS. 11(a) and 11(b) are diagrams for explaining a method of extracting the point cloud data of the left os coxae of a cow. [Figure 12] FIG. 12(a) is a view showing the state of the first partial point cloud data and the second partial point cloud data as viewed from the rear of the cow, and FIG. 12(b) is a view showing the state of the three-dimensional point cloud data as viewed from the side of the cow. [Figure 13] FIG. 13 is a flowchart showing the processing of the evaluation device in the first embodiment. [Figure 14] FIG. 14(a) is a view showing a state where the point cloud data of the right os coxae of a cow is rotated with the plane obtained by moving the second plane by a predetermined distance th as the bottom, FIG. 14(b) is a view of the point cloud data of FIG. 14(a) as viewed from the x direction, and FIG. 14(c) is a view of the point cloud data of FIG. 14(a) as viewed from the y direction. [Figure 15] Figure 15(a) is a schematic diagram showing the shape of the ischium of a cow, and Figure 15(b) is a diagram illustrating the method for calculating the first index value in the first embodiment. [Figure 16] Figure 16(a) shows the relationship between the measured value of the first indicator (r / R of the right ischium) for each sample cow and the body condition score of each cow. Figure 16(b) shows the relationship between the measured value of the second indicator (r / R of the left ischium) for each sample cow and the body condition score of each cow. Figure 16(c) shows the relationship between the average value of the measured first and second indicator values ​​for each sample cow and the body condition score of each cow. [Figure 17] Figure 17 shows an overview of the score table in the first embodiment. [Figure 18] Figure 18(a) shows the point cloud data of the ischial region in a modified example 1 of the first embodiment, and Figure 18(b) shows the point cloud data of Figure 18(a) viewed from the z-axis direction. [Figure 19] Figure 19 shows the relationship between the measured maximum gradient of the ischial bone area for each sample cow and the body condition score of each cow. [Figure 20] Figure 20(a) shows the point cloud data of the ischial region in a modified example 2 of the first embodiment as viewed from the z-axis direction, and Figure 20(b) shows the relationship between the measured minimum gradient of the ischial region of each sample cow and the body condition score of each cow. [Figure 21] Figure 21 shows the results of examining the dimension H and predetermined distance th in modified examples 1 to 3 of the first embodiment. [Figure 22] Figure 22 is a functional block diagram of the focus area estimation device and evaluation device according to the second embodiment. [Figure 23] Figure 23 is a flowchart (part 1) showing the processing of the area of ​​interest estimation device in the second embodiment. [Figure 24] Figure 24 is a flowchart (part 2) showing the processing of the area of ​​interest estimation device in the second embodiment. [Figure 25] Figure 25 shows the first and second partial point cloud data viewed from behind the cow. [Figure 26] Figure 26(a) schematically shows the area where the first partial point cloud data exists (hatched area) as viewed from the right side of the cow, and Figure 26(b) schematically shows the area where the second partial point cloud data exists (hatched area) as viewed from the left side of the cow. [Figure 27] Figure 27 shows an example of the point cloud data near the right hip angle extracted in step S126 of Figure 24 and the point cloud data near the left hip angle extracted in step S128 of Figure 24. [Figure 28] Figure 28 is a diagram illustrating the processes in steps S130 and S132 of Figure 24. [Figure 29] Figure 29 is a flowchart showing the processing of the evaluation device in the second embodiment. [Figure 30] Figure 30(a) shows an example of point cloud data near the right hip angle, and Figure 30(b) shows the point cloud data from Figure 30(a) rotated so that the fourth point is at the highest position. [Figure 31] Figure 31 shows an ellipsoid. [Figure 32] Figure 32 shows the relationship between the measured values ​​of the first and second index values ​​(steepness of the approximated ellipsoid) for each sample cow and the body condition score for each cow in the second embodiment. [Figure 33] Figure 33 shows a cone. [Figure 34] Figure 34 shows the relationship between the measured values ​​of the first and second index values ​​(steepness of the approximate cone) for each sample cow and the body condition score of each cow in a modified example of the second embodiment. [Modes for carrying out the invention]

[0011] 《First Embodiment》 The cattle evaluation system 100 according to the first embodiment will be described below with reference to Figures 1 to 17. Figure 1 shows a schematic representation of the configuration of the cattle evaluation system 100 according to the first embodiment. The cattle evaluation system 100 according to the first embodiment is, as an example, a system for evaluating the body condition score, which indicates the state of body fat accumulation in tethered cattle (for example, dairy cows (Holstein breed)), and as shown in Figure 1, comprises a focus area estimation device 10 as an ischium bone estimation device and an evaluation device 50.

[0012] A three-dimensional point cloud data acquisition device 70 is connected to the focus area estimation device 10. The three-dimensional point cloud data acquisition device 70 is a device capable of acquiring three-dimensional point cloud data of cattle, such as a Kinect sensor (registered trademark), a 3D scanner, or a 360° camera. A rail 72 is installed behind and above the tethered cattle, extending in the direction in which the cattle are lined up, and the three-dimensional point cloud data acquisition device 70 is movable along the rail 72. The three-dimensional point cloud data acquisition device 70 is sequentially positioned near each of the tethered cattle, acquires three-dimensional point cloud data representing the three-dimensional shape of each standing cattle, and transmits it to the cattle evaluation system 100 (focus area estimation device 10). In this first embodiment, the three-dimensional point cloud data acquisition device 70 acquires three-dimensional point cloud data in a range including the left and right ischial bone areas of the cattle (ischial bones and their surroundings on both sides of the tail). Alternatively, an operator holding the three-dimensional point cloud data acquisition device 70 may move behind each cow while acquiring three-dimensional point cloud data of each cow using the three-dimensional point cloud data acquisition device 70.

[0013] Figure 2(a) is a schematic diagram showing an example of the hardware configuration of the area of ​​interest estimation device 10. The area of ​​interest estimation device 10 is an information processing device such as a PC (Personal Computer), and as shown in Figure 2(a), it includes a CPU (Central Processing Unit) 90, ROM (Read Only Memory) 92, RAM (Random Access Memory) 94, storage (HDD (Hard Disk Drive) or SSD (Solid State Drive) etc.) 96, communication unit 97, display unit 93, input unit 95, and portable storage medium drive 99, etc. The display unit 93 includes a liquid crystal display or an organic EL display, and the input unit 95 includes a touch panel, etc. Each of these components of the area of ​​interest estimation device 10 is connected to a bus 98. In the target area estimation device 10, the CPU 90 executes a program (including a program for estimating the ischium of a cow) stored in the ROM 92 or storage 96, or a program read from the portable storage medium 91 by the portable storage medium drive 99, thereby realizing the functions of each part shown in Figure 3. Note that the functions of each part in Figure 3 may also be realized by integrated circuits such as ASICs (Application Specific Integrated Circuits) or FPGAs (Field Programmable Gate Arrays).

[0014] Figure 2(b) is a schematic diagram showing an example of the hardware configuration of the evaluation device 50. The evaluation device 50, like the part of interest estimation device 10, is an information processing device such as a PC, and as shown in Figure 2(b), it is equipped with a CPU 190, ROM 192, RAM 194, storage 196, communication unit 197, display unit 193, input unit 195, and a portable storage medium drive 199, etc. Each of these components of the evaluation device 50 is connected to a bus 198. In the evaluation device 50, the functions of each part shown in Figure 3 are realized by the CPU 190 executing a program (including a cow evaluation program) stored in the ROM 192 or storage 196, or a program read from the portable storage medium 191 by the portable storage medium drive 199. Note that the functions of each part in Figure 3 may be realized by integrated circuits such as ASICs or FPGAs.

[0015] Figure 3 is a functional block diagram of the area of ​​interest estimation device 10 and the evaluation device 50. The functions of the area of ​​interest estimation device 10 and the evaluation device 50 will be described below.

[0016] (Focus area estimation device 10) The area of ​​interest estimation device 10 functions as a point cloud data acquisition unit 12 as an acquisition unit, a tail position identification unit 14 as a first identification unit, a partial point cloud data identification unit 16 as a second identification unit, and an ischial bone extraction unit 18 as an extraction unit, when the CPU 90 executes a program. Figure 3 also shows the cattle position DB 22 stored in the storage 96, etc. The cattle position DB 22 is assumed to manage the identification information of each cattle that is tethered and the location information of the tethered cattle (tethered position information) in association with each other.

[0017] The point cloud data acquisition unit 12 acquires three-dimensional point cloud data representing the three-dimensional shape of a standing cow in three-dimensional space from the three-dimensional point cloud data acquisition device 70. The three-dimensional point cloud data acquisition device 70 detects the position information (data acquisition position information) at the time of acquisition of the three-dimensional point cloud data using a position information detection unit (not shown), and transmits the three-dimensional point cloud data and the data acquisition position information linked together to the area of ​​interest estimation device 10 (point cloud data acquisition unit 12). Therefore, when the point cloud data acquisition unit 12 acquires the three-dimensional point cloud data, it also acquires the data acquisition position information. Furthermore, the point cloud data acquisition unit 12 refers to the cow position DB 22 and identifies which cow the acquired three-dimensional point cloud data belongs to based on the acquired data acquisition position information.

[0018] The tail position identification unit 14 identifies a point (the first point) indicating the position of the cow's tail from the three-dimensional point cloud data acquired by the point cloud data acquisition unit 12.

[0019] The partial point cloud data identification unit 16 identifies point cloud data (first partial point cloud data) that exists in the range including the right ischium of the cow (the ischium located on the right side when viewed from the rear of the cow) based on a point indicating the position of the cow's tail (first point). The partial point cloud data identification unit 16 also identifies point cloud data (second partial point cloud data) that exists in the range including the left ischium of the cow based on the first point.

[0020] The ischial tuberosity extraction unit 18 identifies the point cloud data for the right ischial tuberosity from the first partial point cloud data. The ischial tuberosity extraction unit 18 also identifies the point cloud data for the left ischial tuberosity from the second partial point cloud data. The ischial tuberosity extraction unit 18 outputs the point cloud data for the left and right ischial tuberosities to the evaluation device 50.

[0021] (Evaluation device 50) The evaluation device 50 functions as an index value calculation unit 52 and a BCS evaluation unit 54, as shown in Figure 3, when the CPU 190 executes a program. Figure 3 also shows the score table 60 (see Figure 17) stored in the storage 196, etc.

[0022] The index value calculation unit 52 calculates a first index value from the three-dimensional shape shown by the point cloud data of the right ischium of the cow extracted by the ischium extraction unit 18 of the area of ​​interest estimation device 10. The index value calculation unit 52 also calculates a second index value from the three-dimensional shape shown by the point cloud data of the left ischium of the cow extracted by the ischium extraction unit 18. Details of the first and second index values ​​will be described later.

[0023] The BCS evaluation unit 54 refers to the score table 60 and evaluates the body condition score of the cattle based on the first index value and the second index value calculated by the index value calculation unit 52. The BCS evaluation unit 54 also displays the evaluation results on the display unit 193 or stores them in the storage unit 196.

[0024] (Regarding the processing of the focus area estimation device 10) Next, the processing of the area of ​​interest estimation device 10 will be explained in detail, following the flowchart in Figure 4, and referring to other drawings as appropriate.

[0025] When the process shown in Figure 4 begins, first, in step S10, the point cloud data acquisition unit 12 waits until three-dimensional point cloud data is transmitted from the three-dimensional point cloud data acquisition device 70. Once the three-dimensional point cloud data has been transmitted, the point cloud data acquisition unit 12 proceeds to step S12.

[0026] When the process moves to step S12, the point cloud data acquisition unit 12 acquires the three-dimensional point cloud data transmitted from the three-dimensional point cloud data acquisition device 70, and the data acquisition position information associated with the three-dimensional point cloud data. The three-dimensional point cloud data is as shown in Figures 5(a) and 6(a). Figure 5(a) shows the three-dimensional point cloud data as viewed from the side of the cow, and Figure 6(b) shows the three-dimensional point cloud data as viewed from the rear of the cow. Figure 5(b) shows the area in the three-dimensional image of the cow that corresponds to the three-dimensional point cloud data in Figure 5(a) with a white line frame, and Figure 6(b) shows the area in the three-dimensional image of the cow that corresponds to the three-dimensional point cloud data in Figure 6(a) with a white line frame. In Figures 5(a) and 6(a), the left-right direction of the cow is shown as the x-direction, the front-back direction of the cow is shown as the y-direction, and the up-down direction (vertical direction) of the cow is shown as the z-direction.

[0027] Next, in step S14, the point cloud data acquisition unit 12 refers to the cow position DB 22 and identifies which cow the three-dimensional point cloud data belongs to from the acquired data acquisition position information, and obtains the identification information of that cow. Here, as an example, let's assume that the identification information of cow α has been acquired.

[0028] Next, in step S16, the tail positioning unit 14 approaches the three-dimensional point cloud data from behind and above the cow with the first plane, and identifies the point that first contacts the first plane (the first point). Figure 7 is a diagram illustrating the process in step S16. The tail positioning unit 14 approaches the first plane along a first direction (arrow direction in Figure 7), which is the direction toward the cow from behind and above in three-dimensional space. Here, the first direction is a direction in the yz plane, and the first plane is a plane that intersects (orthogonal in this embodiment) the first direction and is parallel to the X direction. The tail positioning unit 14 also identifies the point in the three-dimensional point cloud data that the first plane first contacts as the first point. Here, the process of moving the first plane in the first direction and identifying the point that the first plane first contacts can also be described as the process of identifying the point that is closest to the first plane, which is set to a position away from the three-dimensional point cloud data in the first direction. Furthermore, since the first point is highly likely to be a point on part of the tail, it can be said to be a point indicating the position of the tail. Here, the angle α of the first plane with respect to the horizontal plane (xy plane) can be any angle between 0° and 90°. However, it is preferable that the angle α is between 20° and 60° in order to increase the likelihood that the first point is a point on part of the tail.

[0029] Next, in step S18, the partial point cloud data identification unit 16 defines an exclusion region as a three-dimensional area having a predetermined dimension in the second direction (x direction), which is the width direction of the cow, and extending in the yz direction with respect to the first point. The partial point cloud data identification unit 16 also identifies the first partial point cloud data that is not included in the exclusion region and is located on one side of the second direction of the exclusion region (the right side when viewed from the rear of the cow).

[0030] Figures 8(a) and 8(b) are diagrams illustrating the process in step S18. Figures 8(a) and 8(b) show the three-dimensional point cloud data as viewed from the rear of the cow. The partial point cloud data identification unit 16 sets the width of the dimension H in the x-direction with the first point as the reference (center), for example, as shown in Figure 8(a). The partial point cloud data identification unit 16 also sets a three-dimensional region having the set width H and extending in the yz direction. Then, as shown in Figure 8(b), the partial point cloud data identification unit 16 sets the set three-dimensional region as an exclusion region, excludes the points included in the exclusion region from the three-dimensional point cloud data, and makes the point cloud data located to the right of the exclusion region the first partial point cloud data. In other words, the three-dimensional point cloud data is divided by the exclusion region, and the point cloud data located to the right of the divided point cloud data is made the first partial point cloud data. This first partial point cloud data includes the point cloud data of the ischium on the right side of the cow. Note that dimension H is assumed to be a value within the range of 4cm to 10cm, as an example.

[0031] Returning to Figure 4, in the next step S20, the partial point cloud data identification unit 16 identifies the second partial point cloud data (see Figure 8(b)) located on the other side of the second direction (x direction) of the excluded region (the left side when viewed from the rear of the cow). That is, it identifies the point cloud data located on the left side of the three-dimensional point cloud data separated by the excluded region as the second partial point cloud data. This second partial point cloud data includes the point cloud data of the left ischium of the cow.

[0032] Next, in step S22, the ischial bone extraction unit 18 approaches the first partial point cloud data with a second plane that intersects (orthogonal in this embodiment) the third direction, which is the direction from the rear and above of the cow toward the cow, and identifies the point (second point) that first contacts the second plane. The ischial bone extraction unit 18 also extracts a portion of the first partial point cloud data that exists within a predetermined range based on the second point as point cloud data of the ischial bone on the right side of the cow. Here, the third direction is a direction in the yz plane and may be the same direction as the first direction in Figure 7. The second plane may also be the same plane as the first plane (a plane inclined at the same angle as the first plane with respect to the horizontal plane).

[0033] Figures 9(a) and 9(b) illustrate a method for extracting point cloud data from the right ischium of a cow. In Figure 9(a), the area where the first partial point cloud data exists is hatched, showing the view from the right side of the cow. In step S22, the ischium extraction unit 18 moves the second plane closer to the first partial point cloud data along the third direction, similar to Figure 7, and identifies the point in the first partial point cloud data that first comes into contact with the second plane as the second point. Figure 10(a) shows the second point as viewed from the rear of the cow. Here, the process of moving the second plane in the third direction and identifying the point that the second plane first comes into contact with can also be described as the process of identifying the point closest to the second plane, which is set to a position away from the three-dimensional point cloud data in the third direction. Next, as shown in Figure 9(b), the ischial bone extraction unit 18 moves the second plane by a predetermined distance th from the second point in the third direction, and extracts the set of points that the second plane passes through during this movement as point cloud data for the right ischial bone of the cow. That is, the ischial bone extraction unit 18 extracts points from the first partial point cloud data that are within the range indicated by hatching in Figure 9(b) as point cloud data for the right ischial bone of the cow. Here, the predetermined distance th is assumed to be a value that falls within the range of 2 to 8 cm. Since the ischial bone of a cow is often about 2 to 8 cm high, it is thought that by setting the predetermined distance th to 2 to 8 cm, the point cloud data for the ischial bone can be extracted with high accuracy.

[0034] Returning to Figure 4, in the next step S24, the ischial bone extraction unit 18 approaches the second partial point cloud data with a third plane that intersects (orthogonal in this embodiment) the fourth direction, which is the direction from the rear and above of the cow toward the cow, and identifies the point (third point) that first contacts the third plane. The ischial bone extraction unit 18 also extracts a portion of the second partial point cloud data that exists within a predetermined range based on the third point as point cloud data of the left ischial bone of the cow. Here, the fourth direction is a direction in the yz plane and may be the same direction as the first direction in Figure 7. The third plane may also be the same plane as the first plane (a plane inclined at the same angle as the first plane with respect to the horizontal plane).

[0035] Figures 11(a) and 11(b) illustrate a method for extracting point cloud data from the left ischium of a cow. In Figure 11(a), the area where the second partial point cloud data exists is hatched, showing the cow as viewed from the left side. The ischium extraction unit 18 approaches the second partial point cloud data from the fourth direction, similar to Figure 9(a), and defines the point where the third plane first makes contact as the third point. Figure 10(b) shows the third point as viewed from the rear of the cow. Next, as shown in Figure 11(b), the ischium extraction unit 18 moves the third plane a predetermined distance th from the third point in the fourth direction, and extracts the set of points that the third plane passes through during this movement as point cloud data for the left ischium of the cow. In other words, the ischial bone extraction unit 18 extracts points located within the area indicated by hatching in Figure 11(b) as point cloud data of the left ischial bone of the cow.

[0036] Next, in step S26, the ischial bone extraction unit 18 links the point cloud data of the left and right ischial bones extracted in steps S22 and S24 with the identification information of the cow identified in step S14 (identification information of cow α) and outputs it to the evaluation device 50. Note that the ischial bone extraction unit 18 may not be able to extract both the point cloud data of the right ischial bone and the point cloud data of the left ischial bone, and may only be able to extract the point cloud data of one of them. In such cases, only the point cloud data that was extracted should be output to the evaluation device 50. Furthermore, if the ischial bone extraction unit 18 is unable to extract both the point cloud data of the right ischial bone and the point cloud data of the left ischial bone, it will not perform the processing in step S26 and will terminate all processing shown in Figure 4.

[0037] With the above steps completed, the entire process shown in Figure 4 by the focus area estimation device 10 is finished.

[0038] Figure 12(a) shows the first and second partial point cloud data viewed from the rear of the cow. In Figure 12(a), each point is shown separately for each part of the lumbar spine, ischium, and spine. Figure 12(b) shows the three-dimensional point cloud data viewed from the side of the cow, with each point also shown separately for each part of the lumbar spine, ischium, and spine.

[0039] Figure 12(a) shows that the spine region (×) can be excluded from the three-dimensional point cloud data by excluding the point cloud data within the exclusion region. It can also be seen that the first partial point cloud data includes the right ischial tuberosity (△), and the second partial point cloud data includes the left ischial tuberosity (△).

[0040] Furthermore, Figure 12(b) shows that the ischial tuberosity (△) is located further back and above the lumbar region (〇) of the cow. Therefore, by excluding the spinal region (×) and performing the processing shown in Figures 9(a), 9(b), and 11(a), 11(b), it is possible to accurately extract the point cloud data of the ischial tuberosity (△) from the first and second partial point cloud data.

[0041] (Regarding the processing of the evaluation device 50) Next, the processing of the evaluation device 50 will be explained following the flowchart in Figure 13, with reference to other drawings.

[0042] When the process shown in Figure 13 begins, first, in step S50, the index value calculation unit 52 waits until the point cloud data of the ischial bone and the cow identification information are transmitted from the area of ​​interest estimation device 10. Once the point cloud data of the ischial bone and the cow identification information are transmitted from the area of ​​interest estimation device 10, the index value calculation unit 52 proceeds to step S52. It is assumed that both the point cloud data of the right ischial bone and the point cloud data of the left ischial bone of the cow are transmitted from the area of ​​interest estimation device 10.

[0043] When the process moves to step S52, the index value calculation unit 52 rotates the point cloud data of the right ischium of the cow, using a plane obtained by moving the second plane by a predetermined distance th as the base, so that the normal vector of the second plane coincides with the z axis. This yields point cloud data as shown in Figure 14(a). The second point in Figure 9 becomes the highest point in the z-axis direction due to this rotation. In Figure 14(a), the point indicated by the square (□) is assumed to be the second point. Note that in Figure 14(a), the origin ((x,y,z)=(0,0,0)) is set at a point (×) located a predetermined distance th to the -z side from the second point. Note that Figure 14(b) is a view of the point cloud data of Figure 14(a) from the x direction, and Figure 14(c) is a view of the point cloud data of Figure 14(a) from the y direction.

[0044] Next, in step S54 of Figure 13, the index value calculation unit 52 calculates a first index value from the three-dimensional shape of the right ischium of the cow (i.e., the point cloud data after rotation as shown in Figure 14(a)).

[0045] Here, the ischium of a cow has a star-shaped three-dimensional form with three vertices, as shown in Figure 15(a). The leaner the cow, the more clearly the ischium is visible and the larger the depression. Conversely, the fatter the cow, the more subcutaneous fat accumulates around the ischium, making the depression smaller and the overall shape closer to a circle. In this first embodiment, the index value calculation unit 52 determines the radius r of the inscribed circle of the ischium (dashed circle in Figure 15(b)) and the radius R of the circumscribed circle of the ischium (dotted circle in Figure 15(b)) from the three-dimensional shape in Figure 14(a). The index value calculation unit 52 then calculates the first index value from the following equation (1). The first index value = r / R …(1)

[0046] Next, in step S56, the index value calculation unit 52 rotates the point cloud data of the left ischium of the cow, using a plane obtained by moving the third plane by a predetermined distance th as the base, so that the normal vector of the third plane coincides with the z axis. As a result, point cloud data similar to that of the right ischium can be obtained, as shown in Figure 14(a).

[0047] Next, in step S58, the index value calculation unit 52 calculates a second index value from the three-dimensional shape of the left ischium of the cow. The second index value is also calculated from the following equation (2) using the radius r of the inscribed circle of the left ischium and the radius R of the circumscribed circle of the left ischium. The second index value = r / R …(2)

[0048] Next, in step S60, the BCS evaluation unit 54 refers to the score table 60 and evaluates the body condition score of the cow from the first index value and the second index value. The body condition score is an index that indicates the state of body fat accumulation. The body condition score is evaluated on a scale of, for example, 1.00 to 5.00, and the higher the value (higher the score), the better the cow is fleshy (fatter).

[0049] Figure 16(a) shows the relationship between the measured first index value (r / R of the right ischium) for each sample cow and the body condition score for each cow as actually evaluated by an expert, plotted on a coordinate system with the first index value on the horizontal axis (x-axis) and the body condition score on the vertical axis (y-axis). Figure 16(a) also shows the approximate straight line derived from the points for each cow. As shown in Figure 16(a), the larger the first index value, the higher the body condition score. In the example in Figure 16(a), the approximate straight line (thick solid line) is y = 2.281x + 1.535, and the coefficient of determination R 2 The result was 0.7614. This confirms a high correlation between the body condition score and the first indicator value.

[0050] Figure 16(b) shows the relationship between the measured second index value (r / R of the left ischium) for each sample cow and the body condition score for each cow as actually evaluated by experts, plotted on a coordinate system with the second index value on the x-axis and the body condition score on the y-axis. As shown in Figure 16(b), the larger the second index value, the higher the body condition score. In the example in Figure 16(b), the approximate straight line (thick solid line) is y = 2.245x + 1.559, and the coefficient of determination R 2 The result was 0.7339. This confirms a high correlation between the body condition score and the second indicator value.

[0051] Figure 16(c) shows the relationship between the average measured values ​​of the first and second indicator values ​​for each sample cow and the body condition score for each cow as actually evaluated by experts. The relationship is plotted on a coordinate system with the average values ​​of the first and second indicator values ​​on the x-axis and the body condition score on the y-axis. As shown in Figure 16(c), the higher the average values ​​of the first and second indicator values, the higher the body condition score. In the example in Figure 16(c), the approximate straight line (thick solid line) is y = 2.281x + 1.535, and the coefficient of determination R 2 The result was 0.7614. This confirms a high correlation between the body condition score and the average values ​​of the first and second indicators.

[0052] Therefore, in this first embodiment, a score table 60 is provided in Figure 3, which stores the input value (x) and the calculation formula for the body condition score (y) in association with each other, as shown in Figure 17. If both the first index value and the second index value can be calculated as a result of processing in steps S52 to S58, the BCS evaluation unit 54 reads formula No. 3 (y = 2.418x + 1.399) from the score table 60 in Figure 17 and calculates the body condition score by substituting the average value (x) of the first index value and the second index value into the formula. If only one of the first index value or the second index value can be obtained as a result of processing in steps S52 to S58, the BCS evaluation unit 54 reads formula No. 1 or No. 2 from the score table 60 in Figure 17 and calculates the body condition score by substituting the first index value or the second index value into the formula.

[0053] Next, in step S62, the BCS evaluation unit 54 displays the body condition score evaluated in step S60 on the display unit 193 and stores it in the storage unit 196. For example, the BCS evaluation unit 54 displays or stores the body condition score along with the cattle identification information (identification information of cattle α) and the evaluation date and time.

[0054] In step S50 of Figure 13, there are cases where only one of the point cloud data for the right ischium of the cow or the point cloud data for the left ischium is transmitted from the area of ​​interest estimation device 10. In such cases, one of the processes in steps S52, S54, S56, or S58 will be omitted.

[0055] With this, all the processes shown in Figure 13 are completed.

[0056] As described in detail above, according to this first embodiment, the point cloud data acquisition unit 12 acquires three-dimensional point cloud data of the cow, and the tail position identification unit 14 approaches the three-dimensional point cloud data from the rear and above of the cow along the first direction, and identifies the first point (a point on part of the tail) that the first plane contacts (Figure 7). Furthermore, the partial point cloud data identification unit 16 excludes a three-dimensional region having a dimension H in the width direction (second direction) of the cow and extending in the yz direction with respect to the first point, and identifies the first partial point cloud data located on one side of the second direction of the exclusion region (right side when viewed from the rear of the cow) and the second partial point cloud data located on the other side of the second direction of the exclusion region (left side when viewed from the rear of the cow) (Figures 8(a), 8(b)). Furthermore, the ischial bone extraction unit 18 approaches the first partial point cloud data with a second plane along a third direction from the rear and above of the cow, identifying a second point that the second plane first contacts (Figure 10(a)), and extracts the set of points within a predetermined range based on the second point as the point cloud data for the right ischial bone (Figure 10(b)). The ischial bone extraction unit 18 also approaches the second partial point cloud data with a third plane along a fourth direction from the rear and above of the cow, identifying a third point that the third plane first contacts (Figure 11(a)), and extracts the set of points within a predetermined range based on the third point as the point cloud data for the left ischial bone (Figure 11(b)). In this way, in this first embodiment, point cloud data corresponding to the left and right ischial bones can be accurately extracted from the three-dimensional point cloud data, taking into account the physical characteristics of the cow.

[0057] Furthermore, in this first embodiment, the ischial bone extraction unit 18 moves the second plane by a predetermined distance th along the third direction from the second point, and extracts a portion of the first partial point cloud data that comes into contact with the second plane during the movement as point cloud data for the right ischial bone. The ischial bone extraction unit 18 also moves the third plane by a predetermined distance th along the fourth direction from the third point, and extracts a portion of the second partial point cloud data that comes into contact with the third plane during the movement as point cloud data for the left ischial bone. This makes it possible to accurately extract point cloud data for the left and right ischial bones, taking into account the position and size of the ischial bones on the cow's body.

[0058] Furthermore, in this first embodiment, the evaluation device 50 evaluates the body condition score from a first index value and a second index value obtained from the three-dimensional shape shown by the point cloud data of the cow's ischium, which has been extracted with high accuracy. This makes it possible to evaluate the body condition score with high accuracy. When a person evaluates the meatiness of a cow, there may be variability in the evaluation depending on the environment in which the evaluation is made and the physical condition of the evaluator, but according to this first embodiment, since the evaluation device 50 evaluates the cow's body condition score, it is possible to suppress variability in the evaluation.

[0059] Furthermore, in this first embodiment, the ratio (r / R) of the diameter (radius r) of the inscribed circle inscribed in the three-dimensional shape of the ischial region and the diameter (radius R) of the circumscribed circle circumscribed around the three-dimensional shape of the ischial region is used as the first and second index values. Since there is a high correlation between the ratio (r / R), which changes depending on the amount of subcutaneous fat, and the body condition score, the body condition score can be evaluated with high accuracy by using the ratio (r / R) to evaluate the body condition score.

[0060] In the first embodiment described above, a predetermined range based on a second point is defined by moving the second plane by a predetermined distance th from the second point in the third direction, and a predetermined range based on a third point is defined by moving the third plane by a predetermined distance th from the third point in the fourth direction. However, the invention is not limited to this. The predetermined range may be defined by other methods. For example, the predetermined range may be defined as the range within a predetermined distance from the second point (or the third point).

[0061] (Variation 1) In the first embodiment described above, the case in which the first and second index values ​​obtained from the three-dimensional shape shown by the point cloud data of the ischial bone of a cow are ratios (r / R) was explained, but the invention is not limited to this. In this modified example 1, information on the maximum gradient of the ischial bone is used as the first and second index values.

[0062] Figure 18(a) shows the point cloud data of the ischial tuberosity. Note that Figure 18(a) shows the point cloud data of the ischial tuberosity rotated by the same process as steps S52 and S56 in Figure 13. In Figure 18(a), the range is divided into predetermined height ranges, and the points included in each range are represented by different types of points.

[0063] Figure 18(b) shows the point cloud data from Figure 18(a) viewed from the z-axis direction. Figure 18(b) shows contour lines indicating the distribution of height.

[0064] In this modified example 1, as shown in Figure 18(b), the index value calculation unit 52 identifies the highest point in the point cloud data (a vertex of the three-dimensional shape, corresponding to the second and third points in the first embodiment). The index value calculation unit 52 also identifies the point (point p) that is closest to the vertex and is included in the lowest height range of the point cloud data. The index value calculation unit 52 then calculates the distance (dmin) between the vertex and point p on the xy plane.

[0065] Furthermore, the index value calculation unit 52 calculates the maximum gradient of the ischial bone from the following equation (3), based on the calculated distance dmin and a predetermined distance th used when extracting the point cloud data of the cow's ischial bone. The predetermined distance th is the height (z-coordinate value) of the vertex of the three-dimensional shape. Maximum gradient of the ischial tuberosity = th / dmin …(3)

[0066] In this modified example 1, the maximum gradient of the right ischial tuberosity is used as the first indicator value, and the maximum gradient of the left ischial tuberosity is used as the second indicator value. A large maximum gradient means that there is little fat around the ischial tuberosity and the shape of the ischial tuberosity is exposed (a lean state), while a small maximum gradient means that there is fat around the ischial tuberosity and the shape of the ischial tuberosity is not easily visible from the outside (a fat state).

[0067] Figure 19 shows the relationship between the measured maximum gradient of the ischial bone area of ​​each sample cow and the body condition score of each cow as actually evaluated by an expert, when the dimension H used to identify the exclusion area is 8 cm and the predetermined distance th is 6 cm, plotted on a coordinate system with the maximum gradient on the horizontal axis (x-axis) and the body condition score on the vertical axis (y-axis). Figure 19 also shows the approximate straight line (dashed line) obtained from the points for each cow. As shown in Figure 19, the larger the maximum gradient, the lower the body condition score. In the example in Figure 19, the approximate straight line (dashed line) is y = -1.3286x + 4.7559, and the coefficient of determination R 2 The result was 0.6567. This confirms a high correlation between the body condition score and the maximum gradient of the ischial tuberosity.

[0068] Therefore, in this modified example 1, the equation of the approximate straight line in Figure 19 is stored in the score table 60 in Figure 3. The BCS evaluation unit 54 then calculates the body condition score by substituting the first index value (maximum slope of the right ischial tuberosity) and the second index value (maximum slope of the left ischial tuberosity) into the equation of the approximate straight line.

[0069] As described above, in this modified example 1, the body condition score is evaluated from the maximum gradient of the ischial tuberosity by utilizing the high correlation between the body condition score and the maximum gradient of the ischial tuberosity. Therefore, similar to the first embodiment, it is possible to evaluate the body condition score with high accuracy.

[0070] (Modification 2) In the above modified example 1, information on the maximum gradient of the ischial tuberosity was used as the first and second indicator values. However, the method is not limited to this, and information on the minimum gradient of the ischial tuberosity may also be used as the first and second indicator values.

[0071] Figure 20(a) shows the point cloud data of the ischial region viewed from the z-axis direction. Figure 20(a) also shows contour lines indicating the distribution of height.

[0072] In this modified example 2, as shown in Figure 20(a), the index value calculation unit 52 identifies the highest point in the point cloud data (a vertex of the three-dimensional shape, corresponding to the second and third points in the first embodiment). The index value calculation unit 52 also identifies the point (point P) that is included in the lowest height range of the point cloud data and is located furthest from the vertex. The index value calculation unit 52 then calculates the distance (Dmax) between the vertex and point P on the xy plane.

[0073] Furthermore, the index value calculation unit 52 calculates the minimum gradient of the ischial bone from the following equation (4), based on the calculated distance Dmax and a predetermined distance th used when extracting the point cloud data of the cow's ischial bone. The predetermined distance th is the height of the vertex (value of the z coordinate). Minimum gradient of the ischial tuberosity = th / Dmax …(4)

[0074] In this modified example 2, the minimum gradient of the right ischial tuberosity is used as the first indicator value, and the minimum gradient of the left ischial tuberosity is used as the second indicator value. A large minimum gradient means that there is little fat around the ischial tuberosity and the shape of the ischial tuberosity is exposed (a thin state), while a small minimum gradient means that there is fat around the ischial tuberosity and the shape of the ischial tuberosity is not easily visible from the outside (a fat state).

[0075] Figure 20(b) shows the relationship between the measured minimum gradient of the ischial bone area of ​​each sample cow and the body condition score of each cow as actually evaluated by an expert, when the dimension H used to identify the exclusion area is 8 cm and the predetermined distance th is 6 cm, plotted on a coordinate system with the minimum gradient on the horizontal axis (x-axis) and the body condition score on the vertical axis (y-axis). Figure 20(b) also shows the approximate straight line (dashed line) obtained from the points for each cow. As shown in Figure 20(b), the larger the minimum gradient, the lower the body condition score. In the example in Figure 20(b), the approximate straight line (dashed line) is y = -5.0416x + 4.6849, and the coefficient of determination R 2 The result was 0.1297. From this, it was determined that there is a correlation between the body condition score and the minimum gradient of the ischial tuberosity.

[0076] Therefore, in this modified example 2, the equation of the approximate straight line in Figure 20(b) is stored in the score table 60 in Figure 3. The BCS evaluation unit 54 then calculates the body condition score by substituting the first index value (minimum slope of the right ischial tuberosity) and the second index value (minimum slope of the left ischial tuberosity) into the equation of the approximate straight line.

[0077] As described above, in this modified example 2, the correlation between the body condition score and the minimum gradient of the ischial tuberosity is utilized to evaluate the body condition score from the minimum gradient of the ischial tuberosity. Therefore, similar to the first embodiment and modified example 1, it is possible to evaluate the body condition score with high accuracy.

[0078] (Variation 3) Furthermore, the values ​​obtained from both the maximum gradient of the ischial tuberosity described in Modification 1 and the minimum gradient of the ischial tuberosity described in Modification 2 may be used as the first and second index values. For example, the average of the maximum and minimum gradients can be used as the first and second index values. In this case, it is thought that the larger the first and second index values, the lower the body condition score will be, and there is a correlation between the body condition score and the first and second index values ​​(the average of the maximum and minimum gradients of the ischial tuberosity).

[0079] In fact, when the inventors verified the correlation by setting the dimension H to 8 cm and the predetermined distance th to 6 cm, the coefficient of determination R 2 The result was 0.674, indicating a high correlation between the body condition score and the first and second index values ​​(average values ​​of the maximum and minimum gradients of the ischial tuberosity).

[0080] Therefore, in this modified example 3, the body condition score is evaluated from the minimum slope of the ischial tuberosity by utilizing the correlation between the body condition score and the average value of the maximum and minimum slopes of the ischial tuberosity. This makes it possible to evaluate the body condition score accurately, as in the first embodiment and modified examples 1 and 2 described above.

[0081] (Regarding the dimension H and the predetermined distance th in modified examples 1 to 3) For the modified examples 1-3, the dimension H is set to five patterns: 2cm, 4cm, 6cm, 8cm, and 10cm, and the predetermined distance th is set to five patterns: 2cm, 4cm, 6cm, 8cm, and 10cm, and the approximate straight line and coefficient of determination R are shown in Figures 19 and 20(b). 2 The process of determining the coefficient of determination was performed for 5 patterns × 5 patterns = 25 patterns. The results are shown in Figure 21. Figure 21 shows the combination of dimension H and predetermined distance th for each pattern, and the coefficient of determination for each of the modified examples 1 to 3.

[0082] In Figure 21, the underlined coefficient of determination value is larger than the others. Therefore, the dimension H should be 7 to 9 cm, preferably 8 cm. Furthermore, the predetermined distance th should be 4 to 10 cm, preferably 4 to 6 cm, in Modification 1, 2 cm, and in Modification 3, 4 to 10 cm, preferably 4 to 8 cm. By determining the dimension H and the predetermined distance th in this way, it becomes possible to accurately evaluate the body condition score.

[0083] 《Second Embodiment》 The second embodiment will be described in detail below with reference to Figures 22 to 32.

[0084] The configuration of the cattle evaluation system 100 in this second embodiment is the same as in Figure 1, and the hardware configuration of the area of ​​interest estimation device 10 and the evaluation device 50, which serve as hip horn estimation devices, is also the same as in Figures 2(a) and 2(b). In this second embodiment, the three-dimensional point cloud data acquisition device 70 acquires three-dimensional point cloud data in a range that includes the left and right hip horns of the cattle (the left and right hip horns and their surroundings located in front of the ischium) and also includes the tail.

[0085] Figure 22 is a functional block diagram of the area of ​​interest estimation device 10 and evaluation device 50 according to the second embodiment.

[0086] (Focus area estimation device 10) As shown in Figure 22, in this second embodiment, the CPU 90 of the area of ​​interest estimation device 10 executes a program (including a program for estimating the hip horn area of ​​a cow), thereby functioning as a point cloud data acquisition unit 12, a tail position identification unit 14, a partial point cloud data identification unit 16, and a hip horn extraction unit 118 as a third identification unit. The point cloud data acquisition unit 12, the tail position identification unit 14, and the partial point cloud data identification unit 16 are the same as in the first embodiment.

[0087] The waist angle extraction unit 118 extracts point cloud data near the right waist angle and point cloud data near the left waist angle from the first partial point cloud data (point cloud data to the right of the excluded area) and the second partial point cloud data (point cloud data to the left of the excluded area), respectively, which have been identified by the partial point cloud data identification unit 16. The waist angle extraction unit 118 also identifies a portion of the right waist angle (the fourth point) from the point cloud data near the right waist angle, and identifies a portion of the left waist angle (the fifth point) from the point cloud data near the left waist angle.

[0088] (Evaluation device 50) In this second embodiment, the CPU 190 of the evaluation device 50 executes a cattle evaluation program, thereby functioning as an index value calculation unit 152 and a BCS evaluation unit 154. In this second embodiment, the index values ​​used when evaluating the body condition score are different from those in the first embodiment and modifications 1 to 3, so the processing contents of the index value calculation unit 152 and the BCS evaluation unit 154 also differ from those of the index value calculation unit 52 and the BCS evaluation unit 54 in the first embodiment. Details of the processing contents of the index value calculation unit 152 and the BCS evaluation unit 154 will be described later.

[0089] (Regarding the processing of the focus area estimation device 10) The processing of the area of ​​interest estimation device 10 in the second embodiment will be described in detail below, following the flowcharts in Figures 23 and 24, and with reference to other drawings as appropriate.

[0090] Once the process shown in Figure 23 begins, steps S10 to S20 are performed in the same manner as in the first embodiment.

[0091] In other words, the point cloud data acquisition unit 12 waits until the three-dimensional point cloud data is transmitted from the three-dimensional point cloud data acquisition device 70 (S10), and when the three-dimensional point cloud data is transmitted, it proceeds to step S12.

[0092] The point cloud data acquisition unit 12 acquires the three-dimensional point cloud data transmitted from the three-dimensional point cloud data acquisition device 70, along with the data acquisition location information associated with the three-dimensional point cloud data (S12). Next, the point cloud data acquisition unit 12 refers to the cow location DB 22 and identifies which cow the three-dimensional point cloud data belongs to based on the acquired data acquisition location information, and acquires the identification information of that cow (S14).

[0093] Next, the tail position identification unit 14 approaches the three-dimensional point cloud data from the rear and above of the cow with the first plane and identifies the point that first contacts the first plane (the first point) (S16, see Figure 7). Next, the partial point cloud data identification unit 16 defines an exclusion region as a three-dimensional area that has a predetermined dimension H in the second direction (x direction), which is the width direction of the cow, and extends in the yz direction with respect to the first point (see Figures 8(a) and 8(b)). The partial point cloud data identification unit 16 also identifies the first partial point cloud data that is not included in the exclusion region and is located on one side of the second direction of the exclusion region (the right side when viewed from the rear of the cow) (S18). Next, the partial point cloud data identification unit 16 identifies the second partial point cloud data that is located on the other side of the second direction (x direction) of the exclusion region (the left side when viewed from the rear of the cow) (S20).

[0094] In this second embodiment, as shown in Figure 25, it is assumed that the first partial point cloud data and the second partial point cloud data have been identified when step S20 is performed. Figure 25 shows the first partial point cloud data and the second partial point cloud data as viewed from behind the cow.

[0095] Next, in step S122, the hip horn extraction unit 118 approaches the first partial point cloud data along a third direction, which is the direction from the rear and above of the cow toward the cow, as shown in Figure 26(a), and identifies the point that first contacts the second plane (the second point). Note that the third direction and the second plane are the same as in the first embodiment, so a detailed explanation is omitted. Figure 26(a) is a schematic diagram showing the area where the first partial point cloud data exists (the hatched area) as viewed from the right side of the cow. Here, the second point can be said to be a point on the right side of the cow's ischium.

[0096] Next, in step S124, the hip horn extraction unit 118 approaches the second partial point cloud data along the fourth direction, which is the direction from the rear and above of the cow toward the cow, as shown in Figure 26(b), and identifies the point that first contacts the third plane (the third point). Note that the fourth direction and the third plane are the same as in the first embodiment, so a detailed explanation is omitted. Figure 26(b) is a schematic diagram showing the area where the second partial point cloud data exists (the hatched area) as viewed from the left side of the cow. Here, the third point can be said to be a point on the left side of the cow's ischium.

[0097] Next, in step S126 of Figure 24, the hip horn extraction unit 118 identifies a position on the anterior side of the cow in three-dimensional space, at an appropriate predetermined distance (450 mm to 750 mm) from the second point identified in step S122. It then deletes points from the first partial point cloud data that are located on the posterior side of the cow beyond the identified position, and extracts the point cloud data on the anterior side of the cow beyond the identified position as "point cloud data near the right hip horn," which corresponds to the part of the cow that includes at least the right hip horn. In this step S126, the point cloud data near the right hip horn is extracted by utilizing the physical characteristics of cows, which indicate that there is a high probability that the hip horn does not exist posterior to a position at a predetermined distance (450 mm to 750 mm) from the second point (a point that is part of the right ischium).

[0098] Next, in step S128, the hip horn extraction unit 118 identifies a position in three-dimensional space that is a predetermined distance (450mm to 750mm) forward of the cow from the third point identified in step S124. It then deletes points from the second partial point cloud data that are located further posterior to the identified position, and extracts the point cloud data located further forward than the identified position as "point cloud data near the left hip horn," which corresponds to the area that includes at least the left hip horn of the cow. In this step S128, the point cloud data near the left hip horn is extracted by utilizing the physical characteristics of cows, which indicate that there is a high probability that the hip horn does not exist posterior to a position at a predetermined distance (450mm to 750mm) from the third point (a point that is part of the left ischium).

[0099] Figure 27 shows an example of point cloud data near the right hip horn extracted in step S126 and point cloud data near the left hip horn extracted in step S128. Figure 27 shows the point cloud data as viewed from the rear of the cow.

[0100] Next, in step S130, the hip horn extraction unit 118 approaches the point cloud data near the right hip horn along the fifth direction, which is the direction from the right side and above the cow toward the cow, as shown in Figure 28, and identifies the point that first contacts the fourth plane (the fourth point). Here, the fifth direction is a direction in the xz plane, and the fourth plane is a plane that intersects (orthogonal in this embodiment) with the fifth direction and is parallel to the y direction (the direction orthogonal to the plane of the paper in Figure 28). Here, the angle β of the fourth plane with respect to the horizontal plane (xy plane) can be any angle between 0° and 90°. It is preferable that the angle β is between 20° and 60° in order to increase the possibility that the fourth point is a point in part of the hip horn.

[0101] Next, in step S132, as shown in Figure 28, the hip horn extraction unit 118 approaches the point cloud data near the left hip horn along the sixth direction, which is the direction from the left side and above toward the cow, and identifies the point that first contacts the fifth plane (the fifth point). Here, the sixth direction is a direction in the xz plane, and the fifth plane is a plane that intersects (orthogonal in this embodiment) with the sixth direction and is parallel to the y direction (the direction orthogonal to the plane of the paper in Figure 28). Here, the angle γ of the fifth plane with respect to the horizontal plane (xy plane) can be any angle between 0° and 90°. It is preferable that the angle γ is between 20° and 60° in order to increase the possibility that the fifth point is a point in part of the hip horn. The angle γ may be the same as or different from the angle β.

[0102] Next, in step S134, the hip horn extraction unit 118 links the point cloud data near the right hip horn, the information of the fourth point, the point cloud data near the left hip horn, and the information of the fifth point with the identification information of the cow identified in step S14, and outputs it to the evaluation device 50.

[0103] With the above steps completed, all processing shown in Figures 23 and 24 by the area of ​​interest estimation device 10 is finished.

[0104] (Regarding the processing of the evaluation device 50) Next, the processing of the evaluation device 50 will be explained following the flowchart in Figure 29, with reference to other drawings.

[0105] When the process shown in Figure 29 begins, in step S150, the index value calculation unit 152 first waits until it receives the following data from the area of ​​interest estimation device 10: point cloud data near the right hip horn, information on the fourth point, point cloud data near the left hip horn, information on the fifth point, and cattle identification information. Once the information is received from the area of ​​interest estimation device 10, the index value calculation unit 152 proceeds to step S152.

[0106] When shifting to step S152, the index value calculation unit 152 rotates the point group data near the right hip angle portion with reference to the fourth point. Specifically, as shown in Fig. 30(a), the point group data near the right hip angle portion is rotated so that the fourth point is at the highest position, resulting in the state shown in Fig. 30(b). Then, the index value calculation unit 152 approximates the shape of the hip angle portion with an ellipsoid.

[0107] An ellipsoid is represented by the following equation (5) and has a shape as shown in Fig. 31. z = c × { (y - y0) 2 / b 2 + (x - x0) 2 / a 2} 1 / 2 …(5)

[0108] Here, as shown in Fig. 31, a is the length of the minor axis of the ellipsoid, b is the length of the major axis, and c represents the height. Also, x0 and y0 are constants indicating the parallel translation amounts for each of the x - axis and y - axis.

[0109] Next, in step S154, the index value calculation unit 152 calculates the first index value from the ellipsoid approximated in step S152. In this second embodiment, the first index value is assumed to be the steepness of the ellipsoid. The steepness of the ellipsoid means the steepness of the hip angle portion. When the steepness of the hip angle portion is large (steep), it means that there is little fat on the hip angle and the cow is thin. On the other hand, when the steepness of the hip angle portion is small (gentle), it means that there is fat on the hip angle and the cow is fat. As an example, the index value calculation unit 152 calculates the steepness of the ellipsoid (the first index value) from the following equation (6) based on the coefficients of the approximated ellipsoid equation. Steepness of the ellipsoid = c / ((a + b) / 2) …(6)

[0110] Next, in step S156, the index value calculation unit 152 rotates the point group data near the left hip angle portion with reference to the fifth point. Then, the index value calculation unit 152 approximates the hip angle portion with an ellipsoid. The processing in this step S154 is the same as the processing in step S152.

[0111] Next, in step S158, the index value calculation unit 152 calculates a second index value from the ellipsoid approximated in step S156. The second index value, like the first index value mentioned above, is the steepness of the ellipsoid, and is obtained from equation (6) above using the values ​​of a, b, and c in the equation of the ellipsoid approximated in step S156.

[0112] Next, in step S160, the BCS evaluation unit 154 refers to the score table 60 and evaluates the body condition score of the cow from the first index value and the second index value.

[0113] Figure 32 shows the relationship between the measured values ​​of the first and second index values ​​(slope steepness of the ellipsoid) for each sample cow and the body condition score for each cow as actually evaluated by experts, plotted on a coordinate system with the first and second index values ​​on the horizontal axis (x-axis) and the body condition score on the vertical axis (y-axis). Figure 32 also shows the approximate straight line derived from the points for each cow. As shown in Figure 32, the steeper the ellipsoid, the lower the body condition score. In the example in Figure 32, the approximate straight line (dashed line) is y = -2.2751x + 4.2978, and the coefficient of determination R 2 The result was 0.7638. This confirms a high correlation between the body condition score and the steepness of the terrain.

[0114] In this second embodiment, the equation of the approximate straight line (y = -2.2751x + 4.2978) is stored in the score table 60 in Figure 22. Therefore, if both the first index value and the second index value can be calculated as a result of processing steps S152 to S158 in Figure 29, the BCS evaluation unit 154 reads the equation from the score table 60 and calculates the body condition score by substituting the average value (x) of the first and second index values ​​into the equation. Alternatively, the BCS evaluation unit 154 may calculate the body condition score by substituting the first index value and the second index value into the equation, and use the average value of each value as the body condition score of the cow. If only one of the first index value or the second index value is obtained as a result of processing steps S152 to S158, the body condition score may be calculated by reading the equation from the score table 60 and substituting the obtained first or second index value into the equation.

[0115] Next, in step S162, the BCS evaluation unit 154 displays the body condition score evaluated in step S160 on the display unit 193 and stores it in the storage unit 196. For example, the BCS evaluation unit 154 displays or stores the body condition score along with the cattle identification information (identification information of cattle α) and the evaluation date and time.

[0116] In step S150 of Figure 29, there are cases where only one of the point cloud data for the right ischium of the cow or the point cloud data for the left ischium is transmitted from the area of ​​interest estimation device 10. In such cases, the evaluation device 50 will omit one of the processes in steps S152, S154 and steps S156, S158.

[0117] With this, all the processes shown in Figure 29 are completed.

[0118] As described in detail above, according to the second embodiment, after identifying the first and second partial point cloud data in the same manner as in the first embodiment (Figure 25), the hip horn extraction unit 118 approaches the first partial point cloud data with a second plane along a third direction from the rear and above of the cow, and identifies the second point to which the second plane first makes contact (Figure 26(a)). At the same time, it approaches the second partial point cloud data with a third plane along a fourth direction from the rear and above of the cow, and identifies the third point to which the third plane first makes contact (Figure 26(b)). Furthermore, the hip horn extraction unit 118 extracts point cloud data from the first partial point cloud data that is located in front of a predetermined distance (450-750 mm) from the second point in front of the cow, and uses this as the point cloud data near the right hip horn (Figure 27). Then, it approaches the point cloud data near the right hip horn with a fourth plane that intersects with the fifth direction, which is on the right side of the cow and moving towards the cow from above, and identifies the fourth point that the fourth plane first makes contact with (Figure 28). In addition, the hip horn extraction unit 118 extracts point cloud data from the second partial point cloud data that is located in front of a predetermined distance from the third point in front of the cow, and uses this as the point cloud data near the left hip horn (Figure 27). Then, it approaches the point cloud data near the left hip horn with a fifth plane that intersects with the sixth direction, which is on the left side of the cow and moving towards the cow from above, and identifies the fifth point that the fifth plane first makes contact with. Thus, in this second embodiment, by taking into account the physical characteristics of the cow, it is possible to accurately extract point cloud data near the right hip horn and point cloud data near the left hip horn, and to identify a fourth point which is a part of the right hip horn and a fifth point which is a part of the left hip horn. Therefore, by using this data, it is possible to accurately estimate the right hip horn and the left hip horn.

[0119] Furthermore, in this second embodiment, based on the data output from the area of ​​interest estimation device 10, the evaluation device 50 approximates the right and left hip horns as ellipsoids, calculates the steepness of the ellipsoids, which are the first and second index values, from the dimensions (a, b, c) of the approximated ellipsoids, and evaluates the body condition score of the cow from the first and second index values. In this second embodiment, the body condition score is evaluated using the steepness of the hip horns, which has a high correlation with the body condition score, making it possible to evaluate the body condition score with high accuracy.

[0120] In the second embodiment described above, the case in which the waist angle extraction unit 118 of the area of ​​interest estimation device 10 identifies a fourth point (a point in part of the right waist angle) and a fifth point (a point in part of the left waist angle) was explained, but the embodiment is not limited to this. For example, the index value calculation unit 152 of the evaluation device 50 may identify the fourth point and the fifth point before step S152 and step S156.

[0121] In the second embodiment described above, the score table 60 may store both an equation for evaluating the body condition score from the first index value and an equation for evaluating the body condition score from the second index value. In this case, the relationship between the measured value of the first index value (steepness of the right hip angle) and the body condition score is plotted on the coordinate system to obtain the equation of the approximate straight line corresponding to the first index value, and the relationship between the measured value of the second index value (steepness of the left hip angle) and the body condition score is plotted on the coordinate system to obtain the equation of the approximate straight line corresponding to the second index value.

[0122] (modified version) In the second embodiment described above, the case in steps S152 and S156 of Figure 29 in which the index value calculation unit 152 approximates the waist angle to an ellipsoid was explained, but it is not limited to this. The index value calculation unit 152 may also approximate the waist angle to a cone. In the following explanation, "ellipsoid" in Figure 29 will be read as "cone".

[0123] Specifically, in step S152, the index value calculation unit 152 rotates the point cloud data near the right side of the hip angle, as shown in Figure 30(a), so that the fourth point is at the highest position, resulting in the state shown in Figure 30(b). Then, the index value calculation unit 152 approximates the hip angle shown in Figure 30(b) as a cone (see Figure 33).

[0124] Here, the cone is represented by the following equation (7) and has the shape shown in Figure 33. z = k - (k / r) * {(y - y0)} 2 +(x-x0) 2} 1 / 2 …(7)

[0125] In equation (7) above, r is the radius of the base, and k is the height of the cone. Also, x0 and y0 are constants that represent the amount of translation along the x and y axes.

[0126] In the next step S154, the index value calculation unit 152 calculates a first index value from the approximated cone. In this modified example, the first index value is assumed to be the steepness of the cone. The steepness of the cone refers to the steepness of the hip angle. A large steepness in the hip angle means that there is little fat in the hip angle and the cow is lean. On the other hand, a small steepness in the hip angle means that there is fat in the hip angle and the cow is fat. As an example, the index value calculation unit 152 calculates the steepness of the cone (first index value) from the following equation (8) based on the values ​​(r, k) obtained from the equation of the approximated cone. The steepness of a cone = k / r …(8)

[0127] Furthermore, in step S156, the index value calculation unit 152 rotates the point cloud data near the left hip angle. Then, the index value calculation unit 152 approximates the left hip angle to a cone. The processing in step S156 is the same as the processing in step S152.

[0128] Furthermore, in step S158, the index value calculation unit 152 calculates a second index value from the cone approximated in step S156. The second index value, like the first index value described above, is the steepness of the cone, and can be obtained from equation (8) above based on the values ​​(r, k) obtained from the equation of the approximated cone.

[0129] Next, in step S160, the BCS evaluation unit 154 refers to the score table 60 and evaluates the body condition score of the cow from the first index value and the second index value.

[0130] Figure 34 shows the relationship between the measured values ​​of the first and second index values ​​(steepness of the approximated cone) for each sample cow and the body condition score for each cow as actually evaluated by experts, plotted on a coordinate system with the first and second index values ​​on the horizontal axis (x-axis) and the body condition score on the vertical axis (y-axis). Figure 34 also shows the approximate straight line derived from the points for each cow. As shown in Figure 34, the steeper the cone, the lower the body condition score. In the example in Figure 34, the approximate straight line (dashed line) is y = -1.9387x + 3.893, and the coefficient of determination R 2 The result was 0.2953. This confirms a correlation between the body condition score and the steepness of the cone.

[0131] In this modified example, the equation of the approximate straight line (y = -1.9387x + 3.893) is stored in the score table 60 in Figure 2. Therefore, if both the first and second indicator values ​​can be calculated as a result of processing steps S152 to S158, the BCS evaluation unit 154 reads the equation from the score table 60 and calculates the body condition score by substituting the average value (x) of the first and second indicator values ​​into the equation. Alternatively, the BCS evaluation unit 154 may calculate the body condition score by substituting each of the first and second indicator values ​​into the equation and use the average value of each value as the body condition score of the cow. If only one of the first or second indicator values ​​can be obtained as a result of processing steps S152 to S158, the body condition score can be calculated by reading the equation from the score table 60 and substituting the obtained first or second indicator value into the equation.

[0132] 《Third Embodiment》 Next, a third embodiment will be described. In this third embodiment, the body condition score of the cow is calculated using both the index value of the ischial bone area described in the first embodiment and the index value of the hip angle area described in the second embodiment.

[0133] For example, if three-dimensional point cloud data of one cow is obtained, the evaluation device 50 performs the processing described in the first embodiment to determine the ratio (=r / R) of the radius r of the inscribed circle of the ischial region to the radius R of the circumscribed circle. The evaluation device 50 also performs the processing described in the second embodiment on the same three-dimensional point cloud data to approximate the hip angle as an ellipsoid and determines the steepness (=c / ((a+b) / 2)) obtained from the ellipsoid. The evaluation device 50 then evaluates the body condition score using both the ratio (=r / R) and the steepness (=c / ((a+b) / 2)).

[0134] Here, the Body Condition Score (BCS) can be expressed by a multiple regression equation as shown in equation (9) below, where the explanatory variables are r / R and c / ((a+b) / 2). BCS=A·(r / R)+B·(c / ((a+b) / 2))+C …(9)

[0135] For example, by examining the relationship between the measured values ​​of the explanatory variables for each sample cow and the body condition scores of each cow as actually evaluated by experts, coefficients A, B, and a constant C were derived, and equation (9) above became equation (9)'. BCS=1.1588·(r / R)-1.1219·(c / ((a+b) / 2))+2.9680 …(9)'

[0136] Note that the coefficient of determination R of the multiple regression equation in equation (9)' above is 2 The correlation coefficient was 0.8422, indicating a high correlation between each explanatory variable and the body condition score.

[0137] Note that equations (9) and (9)' above are just examples. Therefore, other multiple regression equations may be used. For example, a multiple regression equation may be used with the ratio of the right ischial tuberosity (=r / R), the ratio of the left ischial tuberosity (=r / R), the steepness of the right hip angle (=c / ((a+b) / 2)), and the steepness of the left hip angle (=c / ((a+b) / 2)) as explanatory variables.

[0138] As explained above, according to the third embodiment, by obtaining a body condition score from a multiple regression equation with index values ​​obtained from the three-dimensional shape of the ischial tuberosity and index values ​​obtained from the three-dimensional shape of the hip angle as explanatory variables, it is possible to accurately evaluate the body condition score based on the shapes of multiple types of body parts.

[0139] Furthermore, the index value obtained from the three-dimensional shape of the ischial tuberosity does not have to be a ratio (=r / R), but may be the index value shown in Modifications 1 to 3 of the first embodiment. Also, the index value obtained from the three-dimensional shape of the hip angle does not have to be the steepness of an approximated ellipsoid (=c / ((a+b) / 2)), but may be the steepness of a cone shown in the Modification of the second embodiment.

[0140] In the above embodiments and modifications, the BCS evaluation units 54 and 154 have been described as using the equation of the approximate straight line or the multiple regression equation stored in the score table 60 to determine the body condition score from the index values, but this is not the only way. The BCS evaluation units 54 and 154 may also prepare a large amount of training data combining each index value and the body condition score evaluated by an expert, and evaluate the body condition score by inputting the index values ​​calculated by the index value calculation units 52 and 152 into a system that has been trained on this training data.

[0141] In the above embodiments and modifications, the case in which the point cloud data acquisition unit 12 identifies which cow the three-dimensional point cloud data belongs to by referring to the cow position DB 22 has been described, but this is not the only case. For example, if a barcode or characters indicating the cow's identification information are provided on or near the cow, the three-dimensional point cloud data acquisition device 70 may read the barcode or characters when acquiring the three-dimensional point cloud data to identify the cow's identification information.

[0142] In the above embodiments and modifications, the case where the area of ​​interest estimation device 10 and the evaluation device 50 are separate devices has been described. However, the functions of the area of ​​interest estimation device 10 and the evaluation device 50 may be implemented in a single device. Also, in the above embodiments and modifications, the case where the three-dimensional point cloud data acquisition device 70 and the cattle evaluation system 100 are separate devices has been described. However, the functions of the three-dimensional point cloud data acquisition device 70 and the cattle evaluation system 100 may be implemented in a single device. In this case, for example, if the functions of the three-dimensional point cloud data acquisition device 70 are provided by a portable information device such as a smartphone, the functions of the area of ​​interest estimation device 10 and the evaluation device 50 may be provided by the portable information device. Furthermore, some of the functions of the area of ​​interest estimation device 10 and the evaluation device 50 may be provided by an external device (server).

[0143] In the embodiments and modifications described above, the body condition score was used as an example to evaluate the meatiness of a cow. However, other indicators of meatiness may also be used. Furthermore, the evaluation is not limited to dairy cows; it can also be applied to beef cattle and other types of cattle.

[0144] In the above embodiments and modifications, the case in which the cattle evaluation system 100 is used in a barn where cattle are kept tethered, as shown in Figure 1, has been described, but it is not limited to this. For example, it can also be used in free-stall barns or free-barn barns. In this case, when the cattle enter the facility dedicated to milking or when milking is being performed inside the facility, the three-dimensional point cloud data acquisition device 70 installed in the facility may acquire three-dimensional point cloud data of the cattle. Alternatively, when the cattle are lined up eating feed, the three-dimensional point cloud data of the cattle may be acquired by the three-dimensional point cloud data acquisition device 70 moving along the rail 72, as in Figure 1.

[0145] Although embodiments of the present invention have been described in detail above, the present invention is not limited to these specific embodiments, and various modifications and changes are possible within the scope of the gist of the present invention as described in the claims. In the above description, numerical ranges expressed using "~" include the lower limit and upper limit of the numerical value itself. [Explanation of Symbols]

[0146] 10. Target area estimation device (ischial tuberosity estimation device, lumbar angle estimation device) 12. Point cloud data acquisition unit (acquisition unit) 14. Tail position identification unit (first identification unit) 16. Partial point cloud data identification unit (second identification unit) 18. Ischial region extraction section (extraction section) 22 Cow position DB 50 Evaluation device 52. Indicator Value Calculation Section 54 BCS Evaluation Department 60 Score Table 90 CPU (Computer) 100 Cattle Evaluation System 118 Waist corner extraction part (3rd specific part) 152 Indicator Value Calculation Section 154 BCS Evaluation Department

Claims

1. An acquisition unit that acquires point cloud data representing the three-dimensional shape of a cow in a standing position in three-dimensional space, In the three-dimensional space, a first identification unit brings a first plane intersecting the first direction closer to the point cloud data along a first direction which is the direction toward the cow from behind and above the cow, and identifies a first point which the first plane first contacts. In the aforementioned three-dimensional space, a three-dimensional region having a predetermined dimension in the second direction, which is the width direction of the cow, and extending along the first direction with respect to the first point, is defined as an exclusion region, and a second identification unit identifies a first partial point cloud data that is located on one side of the exclusion region in the second direction among the point cloud data not included in the exclusion region, In the three-dimensional space, an extraction unit brings a second plane intersecting the third direction closer to the first partial point cloud data along a third direction which is the direction toward the cow from behind and above, identifies a second point that the second plane first contacts, and extracts a portion of the first partial point cloud data that exists within a predetermined range based on the second point as point cloud data of one ischium of the cow. A device for estimating the ischial bone location of a cow.

2. The cattle ischium estimation device according to claim 1, characterized in that the extraction unit identifies the second point, moves the second plane a predetermined distance along the third direction from the second point, and extracts a portion of the first partial point cloud data that comes into contact with the second plane during the movement as point cloud data of one ischium of the cattle.

3. The second identification unit identifies a second partial point cloud data that is located on the other side of the second direction of the exclusion region among the point cloud data that is not included in the exclusion region, The extraction unit moves a third plane intersecting the fourth direction towards the cow from the rear and above in the three-dimensional space, and brings it closer to the second partial point cloud data. It then extracts a portion of the second partial point cloud data that exists within a predetermined range based on the third point that the third plane first contacts, as point cloud data for the other ischium of the cow. The cattle ischium estimation device according to feature 1.

4. The cattle ischium estimation device according to claim 3, characterized in that the extraction unit identifies the third point, moves the third plane a predetermined distance along the fourth direction from the third point, and extracts a portion of the second partial point cloud data that comes into contact with the third plane during the movement as point cloud data of the other ischium of the cattle.

5. A device for estimating the ischial bone of a cow according to any one of claims 1 to 4, A cattle evaluation system comprising: an evaluation device for evaluating the meatiness of a cattle from index values ​​obtained from the three-dimensional shape shown by the point cloud data of the cattle's ischial bone extracted by the cattle's ischial bone estimation device.

6. The cattle evaluation system according to claim 5, characterized in that the index value is the ratio of the diameter of the inscribed circle inscribed in the three-dimensional shape of the ischial bone to the diameter of the circumscribed circle circumscribed around the three-dimensional shape of the ischial bone.

7. The cattle evaluation system according to claim 5, characterized in that the index value is one of the maximum gradient value, the minimum gradient value, or a value obtained from the maximum gradient value and the minimum gradient value in the three-dimensional shape of the ischium.

8. The cattle evaluation system according to claim 5, wherein the evaluation device evaluates the body condition score as an evaluation of the meatiness of the cattle.

9. The cattle evaluation system according to claim 8, characterized in that the evaluation device evaluates the body condition score based on an equation showing the relationship between the body condition score and the index value.

10. We obtained point cloud data representing the three-dimensional shape of a standing cow in three-dimensional space. In the three-dimensional space, a first plane intersecting the first direction is brought closer to the point cloud data along a first direction which is the direction toward the cow from behind and above the cow, and a first point which the first plane first contacts is identified. In the aforementioned three-dimensional space, a three-dimensional region having a predetermined dimension in the second direction, which is the width direction of the cow, and extending along the first direction with respect to the first point, is defined as an exclusion region, and a first partial point cloud data existing on one side of the exclusion region in the second direction is identified from the point cloud data not included in the exclusion region. In the three-dimensional space, a second plane intersecting the third direction is brought closer to the first partial point cloud data along a third direction which is the direction toward the cow from behind and above the cow, the second point that the second plane first contacts is identified, and a portion of the first partial point cloud data that exists within a predetermined range based on the second point is extracted as point cloud data for one of the cow's ischial bones. A program for estimating the ischial bone of a cow, characterized by having a computer perform the processing.

Images