Analysis device, training device, and program

Abstract

Provided is an analysis device that comprises a load data acquisition unit that acquires measurement data in association with identification information for respective individuals, the measurement data being from a plurality of sensor members that measure load and are provided along a path that is walked by living bodies, an image data acquisition unit that acquires images captured of the living bodies that walk the path and the identification information, a location detection unit that generates location data that indicates whether the legs of the living bodies have pressed any of the plurality of sensor members on the basis of the images, and a state estimation unit that estimates the states of the living bodies on the basis of the measurement data and the location data.

Description

Analysis device, training device and program

[0001] The present invention relates to an analysis device, a training device, and a program.

[0002] Patent Document 1 discloses a technology for "objectively diagnosing the condition of the hoof by capturing the movement of a specific part of a dairy cow in a captured image and diagnosing the condition of the hoof of the captured dairy cow based on the movement of this specific part (Abstract)." Patent Document 2 discloses a technology for "calculating a score indicating the health condition of a cow based on a group of three-dimensional coordinates representing the three-dimensional shape of the cow extracted from a range image of the cow (Abstract)." Non-Patent Document 1 discloses a technology for early detection of lameness in cows. Patent Document 1: JP 2005-253435 A Patent Document 2: WO 2017 / 187719 Non-Patent Document 1: A. Van Nuffel, et al., "Exploration of measurement variation of gait variables for early lameness detection in cattle using the GAITWISE," Livestock Science, Volume 156, Issues 1-3, September 2013, Pages 88-95. General disclosure

[0003] A first aspect of the present invention provides an analysis device. The analysis device may include a load data acquisition unit that acquires measurement data from multiple sensor elements that measure loads and are arranged on a path along which a living organism walks, in association with identification information for each individual. Any of the analysis devices may include an image data acquisition unit that acquires images of the living organism along the path along with the identification information. Any of the analysis devices may include a position detection unit that generates, based on the images, position data indicating which of the multiple sensor elements each leg of the living organism presses. Any of the analysis devices may include a state estimation unit that estimates the state of the living organism based on the measurement data and the position data.

[0004] In any of the above analysis devices, the position detection unit may determine whether a ground contact portion of any leg of the living body straddles two or more of the sensor members. In any of the above analysis devices, when the ground contact portion straddles two or more of the sensor members, the state estimation unit may calculate a measurement value of the load of the ground contact portion based on measurement values ​​of the loads of the two or more sensor members.

[0005] In any one of the above analytical devices, each of the plurality of sensor elements may be in the form of a plate having a pressing surface, and the length and width of the pressing surface of each of the sensor elements may be 20 cm or more and 50 cm or less.

[0006] In any of the above analysis devices, each of the plurality of sensor members may be plate-shaped and have a pressing surface. In any of the above analysis devices, each of the sensor members may measure a load distribution on the pressing surface. In any of the above analysis devices, the state estimation unit may estimate the state of the living body based on the load distribution on the ground-contact portions of each leg of the living body.

[0007] Any of the above analysis devices may include a speed detection unit that detects a moving speed of the living body based on the image. In any of the above analysis devices, the state estimation unit may estimate the state of the living body further based on the moving speed of the living body.

[0008] In a second aspect of the present invention, there is provided an analysis device. The analysis device may include a load data acquisition unit that is arranged on a path along which a living organism walks and acquires measurement data from a sensor member that measures the distribution of loads on the ground contact portions of each leg of the living organism, in association with identification information for each individual. Any of the analysis devices may include a state estimation unit that estimates the state of the living organism based on the distribution of loads on the ground contact portions of each leg of the living organism.

[0009] In any of the above analysis devices, the condition estimation unit may estimate a type of hoof disease of the living organism.

[0010] In any of the above analysis devices, the state estimation unit may estimate the state of the living organism based on at least one feature of uneven load on each leg, difference from past measurement data of the same individual, and difference from the measurement data of other individuals.

[0011] Any of the above analysis devices may include a weight calculation unit that calculates the weight of the living body based on the measurement data.

[0012] In any of the above analysis devices, the state estimation unit may select an estimation method for estimating the state of the living organism based on an estimation result of whether the living organism is walking or stationary.

[0013] A third aspect of the present invention provides a training device. The training device includes a training data acquisition unit that acquires training data including a load distribution at the ground contact portion of each leg of a living organism in association with identification information of each individual. Any of the training devices may include an evaluation unit that inputs the training data to a training subject model and acquires an output result indicating the state of the living organism from the training subject model. Any of the training devices may include a training unit that adjusts the training subject model based on label data indicating the state of the individual corresponding to the identification information and the output result.

[0014] In any of the above training devices, the training subject model may output the type of hoof disease of the living organism as the state of the living organism.

[0015] Any of the above training devices may include a data processing unit that estimates which leg of the living body the load distribution included in the training data corresponds to and includes the estimation result in the training data. In any of the above training devices, the evaluation unit may input the training data including the estimation result to the training target model.

[0016] In any of the above training devices, the data processing unit may further perform at least one of noise removal and upsampling or downsampling of the data set in space or time series.

[0017] Any of the above training devices may include a data processing unit that estimates which leg of the living body the load distribution included in the training data corresponds to and includes the estimation result in the training data. In any of the above training devices, the evaluation unit may input the training data including the estimation result to the training target model.

[0018] In a fourth aspect of the present invention, there is provided a program for causing a computer to function as the analysis device according to claim 1 or 6.

[0019] It should be noted that the above summary does not list all of the necessary features of the present invention, and that subcombinations of these features may also constitute inventions.

[0020] 1 is a block diagram showing an example of an analysis device 100 according to an embodiment of the present invention. FIG. 1 is a schematic diagram illustrating an example of a sensor unit 110. FIG. 1 shows an example of a plurality of sensor members 112 viewed from above. FIG. 2 is a diagram showing an example of measurement results of loads on each leg of the same individual. FIG. 3 is a diagram showing an example of measurement results of loads on a plurality of individuals. FIG. 4 shows an example of a time waveform of measurement results output by one sensor member 112. FIG. 5 is a diagram showing another example of measurement results of loads in one sensor member 112. FIG. 6 is a diagram showing an example of the position of the center of gravity of load in the contact portion 211. FIG. 7 is a diagram showing another example of the configuration of the analysis device 100. FIG. 8 is a block diagram showing an example of a training device 300 that trains a training subject model. FIG. 9 shows an example of a computer 1200 in which multiple aspects of the present invention may be embodied in whole or in part.

[0021] The present invention will be described below through embodiments of the invention, but the following embodiments do not limit the scope of the invention as claimed. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention.

[0022] FIG. 1 is a block diagram illustrating an example of an analysis device 100 according to an embodiment of the present invention. The analysis device 100 estimates the state of a living organism based on load measurement data obtained by sensor elements arranged along the walking path of the living organism. The state of a living organism refers to a health-related condition of the living organism, such as illness or injury. The state of a living organism may be a condition indicating whether the living organism is currently ill or injured, or a condition indicating a precursor to illness or injury. The living organism may be an animal such as a cow, or may be a human. The state of a living organism may be a state in which the animal is lame, or a condition indicating a precursor to lameness. If the living organism is an animal with hooves, the state of the living organism may be the state of the hooves. By using the load measurement data, abnormalities in the state of a living organism can be detected early. For example, abnormalities in the state of a living organism can be detected early, even before changes in the appearance of the living organism are apparent.

[0023] Analysis device 100 includes a load data acquisition unit 10 and a state estimation unit 50. Analysis device 100 may further include an image data acquisition unit 20 and a position detection unit 30. Analysis device 100 may further include at least one of a sensor unit 110 and an imaging unit 120. Sensor unit 110 and imaging unit 120 may be provided outside analysis device 100. For example, analysis device 100 may be a computer, and sensor unit 110 and imaging unit 120 may be external devices that communicate with analysis device 100. A program for causing the computer to function as analysis device 100 may be installed in the computer.

[0024] The sensor unit 110 is placed on a path along which the living organism walks and measures the load of the living organism. The sensor unit 110 may measure the load of a living organism that is walking, or may measure the load of a living organism that is stationary and not walking. The sensor unit 110 includes one or more sensor elements. Each sensor element may be in the shape of a plate having a pressure surface. Each sensor element measures the load applied to the pressure surface. One sensor element may output one measurement value. The measurement value may be the sum of the loads applied to the pressure surface. In another example, each sensor element may output the distribution of the load on one pressure surface.

[0025] The imaging unit 120 captures an image of a living organism along a path on which the sensor unit 110 is provided. The imaging unit 120 may capture an image of a living organism that is walking, or may capture an image of a living organism that is stationary and not walking. The imaging unit 120 captures an image that makes it possible to determine at least which sensor element is pressed by each leg of the living organism. The imaging unit 120 may capture an image including the pressed surface of each sensor element. The imaging unit 120 may capture a video of the living organism. In another example, the imaging unit 120 may capture a still image of the living organism including any leg of the living organism each time that leg comes into contact with a sensor element. Only one imaging unit 120 may be provided, or multiple imaging units 120 may be provided.

[0026] The load data acquiring unit 10 acquires the measurement data from the sensor unit 110 in association with the identification information of each individual. The identification information may be acquired by the sensor unit 110 and notified to the load data acquiring unit 10 together with the measurement data. The measurement data may include a timestamp indicating the date and time when the load was measured.

[0027] The image data acquisition unit 20 acquires image data captured by the imaging unit 120 in association with identification information of each individual. The identification information may be acquired by the imaging unit 120 and notified to the image data acquisition unit 20 together with the image data. The image data may include a timestamp indicating the date and time the image was captured.

[0028] The position detection unit 30 generates position data indicating the position at which each leg of the living organism presses the sensor unit 110, based on the image data acquired by the image data acquisition unit 20. The position detection unit 30 notifies the state estimation unit 50 of the position data indicating the detected position. The position data may include data indicating which sensor member each leg of the living organism presses. The position data may also include data indicating which position on the pressing surface of each sensor member each leg of the living organism presses.

[0029] The state estimation unit 50 estimates the state of the living organism based on the measurement data. The state estimation unit 50 may estimate the state of the living organism based further on the position data. The state estimation unit 50 may calculate the load on each leg of the living organism from the measurement data and the position data. For example, if the living organism is a cow, the state estimation unit 50 may calculate the magnitude of the load on each of the cow's left front leg, left hind leg, right front leg, and right hind leg. The load on each leg is expressed as a value per unit area (for example, N / m ２ ) or the total load (N) at the ground contact portion of the leg. The magnitude of the load on each leg may be the maximum value of the load during the period from when the leg starts pressing against the sensor member to when it stops (or the period from when the leg contacts the sensor member to when it separates). The state estimation unit 50 may also estimate whether the living organism is walking or stationary based on the measurement data or position data. A state in which the living organism is standing and not walking may be included in the stationary state. The state estimation unit 50 may calculate the walking speed based on the measurement data or position data, and determine that the living organism is standing or stationary when the walking speed is below a threshold. The state estimation unit 50 may also switch algorithms, such as machine learning models, used to estimate abnormalities depending on the estimation result of the walking state, standing, or stationary state. An abnormality estimation algorithm corresponding to the walking state and an abnormality estimation algorithm corresponding to the standing or stationary state may be preset in the state estimation unit 50.

[0030] In a healthy living organism, the load on each leg tends to be uniform when walking or standing still. The state estimation unit 50 may estimate the state of the living organism based on the variation in the load on each leg. For example, the state estimation unit 50 may determine that the state of one leg is abnormal when the difference between the load on one leg and the average load on the other legs is equal to or greater than a predetermined value. However, the method of determining the state is not limited to the method described above.

[0031] Fig. 2 is a schematic diagram illustrating an example of the sensor unit 110. Fig. 2 shows a side view of a walking living body 200. In this specification and the drawings, the axis parallel to the direction of gravity is referred to as the Z axis, the axis parallel to the moving direction of the living body 200 is referred to as the X axis, and the axis perpendicular to both the X axis and the Y axis is referred to as the Y axis.

[0032] The sensor unit 110 of this example includes multiple sensor elements 112. In this example, each sensor element 112 has a plate shape with a pressure surface 116. The pressure surface 116 is a surface capable of detecting applied pressure. The pressure surface 116 may be the surface with the largest area among the surfaces of the sensor element 112. Each sensor element 112 is arranged so that the pressure surface 116 faces upward in the Z-axis direction. The multiple sensor elements 112 are arranged on a path 114 along which the living organism 200 walks, along the direction of travel of the living organism 200. The pressure surfaces 116 of adjacent sensor elements 112 may be in contact with each other or may be slightly spaced apart. The distance between adjacent pressure surfaces 116 is preferably shorter than the maximum length of the bottom surface of the living organism 200's leg. The distance between the pressure surfaces 116 may be less than half, or even less than one-quarter, of the maximum length of the bottom surface of the leg. This distance is, for example, 10 cm or less.

[0033] The path 114 may be, for example, an aisle in the stable of the living organism 200 or the floor of a milking area, but is not limited to these. The multiple sensor members 112 may also be arranged in a direction (Y-axis direction) perpendicular to the direction of travel (X-axis direction). As an example, two sensor members 112 are arranged side by side in the Y-axis direction. This makes it possible to measure the load on the right leg and the load on the left leg of the living organism 200 independently.

[0034] The imaging unit 120 captures an image that allows determination of which sensor member 112 is pressed by each leg of the living body 200. Although one imaging unit 120 is shown in Fig. 2, images of the legs of the living body 200 may be captured by a plurality of imaging units 120.

[0035] In this example, the living body 200 is provided with a tag 202 that transmits identification information of the individual. The tag 202 may be an RFID tag that transmits the identification information. The image capturing unit 120 and the sensor unit 110 may receive the identification information from the RFID tag and associate the image data and measurement data with the identification information.

[0036] 3 shows an example of multiple sensor members 112 viewed from above. In this example, sensor members 112-1 to 112-5 are lined up on the right side of the living body 200 in the direction of travel, and sensor members 112-6 to 112-10 are lined up on the left side. FIG. 3 also shows ground contact portions 211, 212, 213, and 214 where each leg of the living body 200 comes into contact with the sensor members 112. As an example, ground contact portion 211 corresponds to the left hind leg, ground contact portion 212 corresponds to the left front leg, ground contact portion 213 corresponds to the right hind leg, and ground contact portion 214 corresponds to the right front leg. For ground contact portions 211 and 212, the positions after the legs have been moved are indicated by dashed lines.

[0037] The imaging unit 120 captures images that allow the positions of the ground contact portions 211, 212, 213, and 214 of each sensor member 112 to be determined. This makes it possible to measure the load on each leg of the living body 200. In this example, two imaging units 120-1 and 120-2 are provided. The imaging unit 120-1 captures images of the sensor members 112-1 to 112-5 on the right side, and the imaging unit 120-2 captures images of the sensor members 112-6 to 112-10 on the left side. The imaging units 120-1 and 120-2 differ in at least one of the imaging position and the imaging angle. An RGB camera, LiDAR, 3D camera, or the like can be used as the imaging unit 120.

[0038] The position detection unit 30 detects the positions of the ground contact portions 211 to 214. The position detection unit 30 may determine whether or not the ground contact portions 211 to 214 of any of the legs straddle the pressing surfaces 116 of two or more sensor members 112. In the example of Fig. 3, the ground contact portion 213 straddles two sensor members 112-1 and 112-2 that are adjacent to each other in the X-axis direction. Similarly, the ground contact portion 212 straddles two adjacent sensor members 112.

[0039] When any of the ground contact portions 211 to 214 straddles two or more sensor members 112, the state estimation unit 50 calculates the measurement value of the load of the ground contact portion based on the measurement values ​​of the loads of the two or more sensor members 112. For example, the state estimation unit 50 calculates the measurement value of the load of the ground contact portion 213 based on the measurement value of the load of the sensor member 112-1 and the measurement value of the load of the sensor member 112-2. The state estimation unit 50 may calculate the sum of the measurement value of the load of the sensor member 112-1 and the measurement value of the load of the sensor member 112-2 as the load of the ground contact portion 213. When any of the ground contact portions 211 to 214 does not straddle two or more sensor members 112, the state estimation unit 50 calculates the measurement value of the load of one ground contact portion based on the measurement value of the load of one sensor member 112.

[0040] The sensor unit 110 of this example has multiple sensor elements 112 arranged along a path 114 along which the living body 200 walks. This configuration allows the use of relatively small sensor elements 112. This makes it easy to arrange multiple sensor elements 112 along the path 114 even if the path 114 is not straight. It also reduces the cost of the sensor elements 112. On the other hand, when multiple sensor elements 112 are arranged and used, the ground contact portion of the living body 200's leg may straddle multiple sensor elements 112. In this example, by analyzing image data from the imaging unit 120, it is possible to easily detect which sensor element 112 the ground contact portion straddles. Therefore, even when multiple sensor elements 112 are used, the load on each leg of the living body 200 can be measured with high accuracy.

[0041] The position detection unit 30 may determine whether the pressing surface 116 of any sensor member 112 is being pressed by two or more legs simultaneously. The state estimation unit 50 may process the measurement value of a sensor member 112 that is being pressed by two or more legs simultaneously as an invalid value.

[0042] The position detection unit 30 may detect the position of the ground contact portion within the plane of the pressing surface 116 of each sensor member 112. The state estimation unit 50 may set a correction value for each sensor member 112 according to the position of the ground contact portion within the plane of the pressing surface 116. The state estimation unit 50 may correct the measured value of the load of the sensor member 112 using the correction value according to the position of the ground contact portion detected by the position detection unit 30. With this configuration, the load on each leg of the living body 200 can be measured with high accuracy even for sensor members 112 whose sensitivity to load varies depending on the position within the plane of the pressing surface 116.

[0043] The position detection unit 30 may detect the position of the ground contact portion based on shape information set in advance regarding the shape of the ground contact portion of the living body 200 to be measured. The position detection unit 30 may detect the area where pressure is detected as the ground contact portion when the shape of the area within the pressure surface 116 matches the shape information. The position detection unit 30 may detect the position of the ground contact portion using both image data and shape information. For example, when the position of the ground contact portion detected from the image data matches the position of the ground contact portion detected based on the shape information, the position detection unit 30 may adopt the position as the position of the ground contact portion.

[0044] The position detection unit 30 may detect the tip of a leg, such as a hoof, from an image of the living organism 200 to be measured, based on a leg detection model generated by machine learning. For example, the leg detection model may be a model that extracts a hoof from an image of the living organism 200 to be measured by learning in advance images of the hooves of a plurality of individuals. The leg detection model may be a model that extracts the tip of a leg from an image of the living organism 200 to be measured by learning in advance images of the legs of a plurality of individuals.

[0045] The position detection unit 30 may detect the position of the ground contact portion based on acceleration information indicating the acceleration of the legs or hooves of the living organism 200 being measured. The acceleration information may be measured by an acceleration sensor attached to each leg or each hoove of the living organism 200. The position detection unit 30 may detect the timing when the legs or hooves of the living organism 200 touch the ground based on the acceleration information. For example, the position detection unit 30 may detect the timing when the acceleration of the legs or hooves of the living organism 200 reaches 0 or a minimum value as the ground contact timing.

[0046] The position of each leg or hoof of the living organism 200 at a predetermined start timing is set as the initial position of each leg or hoof. The position detection unit 30 may calculate the movement speed of each leg or hoof of the living organism 200 by integrating, with time, the acceleration during a target period from the start timing to the ground contact timing at which the acceleration is 0 or a minimum value. In this case, the position detection unit 30 acquires acceleration information of each leg or hoof in the movement direction of the living organism 200. The position detection unit 30 may calculate the movement distance of each leg or hoof of the living organism 200 by integrating, with time, the movement speed of each leg or hoof of the living organism 200 during the target period. The position detection unit 30 may calculate a position that is the movement distance away from the initial position of each leg or hoof of the living organism 200 in the movement direction as the position of the ground contact portion of the living organism 200.

[0047] The position detection unit 30 may acquire acceleration information in multiple directions. For example, the position detection unit 30 may acquire acceleration information in two directions in a plane parallel to the ground. In this case, the in-plane position of each leg or hoof of the living organism 200 can be calculated from the acceleration information. The position detection unit 30 may acquire acceleration information in each direction of three mutually perpendicular axes. From the acceleration information, the position of each leg or hoof of the living organism 200 in three-dimensional space can be calculated.

[0048] In another example, the position where the sensor member 112 is pressed at the ground contact timing calculated from the acceleration information may be detected as the position of the ground contact part of each leg of the living body 200. In this case, the position detection unit 30 detects the position of the ground contact part from the acceleration information and the measurement data.

[0049] The position detection unit 30 may detect the position of the ground contact portion using both the image data and the acceleration information. For example, when the position of the ground contact portion detected from the image data and the position of the ground contact portion detected based on the acceleration information match, the position detection unit 30 may adopt the position as the position of the ground contact portion. The position detection unit 30 may also detect the position of the ground contact portion using the shape information described above.

[0050] As described above, the position detection unit 30 may detect the position of the ground contact portion based only on the image data of the living body 200 and the measurement data of the load. The position detection unit 30 may detect the position of the ground contact portion using both the image data of the living body 200 and the measurement data of the load. The position detection unit 30 may detect the position of the ground contact portion based only on the measurement data of the image data of the living body 200 and the measurement data of the load.

[0051] The position detection unit 30 may generate position data from image data, etc., when set conditions are satisfied. For example, the position detection unit 30 may generate position data from image data when the magnitude of the load measured by the sensor unit 110 is equal to or greater than a predetermined value. Similarly, the imaging unit 120 and the image data acquisition unit 20 may acquire and output image data when set conditions are satisfied.

[0052] The position detection unit 30 may generate position data when the image data captured by the imaging unit 120 satisfies a set condition. For example, the position detection unit 30 may generate position data from the image data when the fluctuation in the luminance value of a predetermined pixel in the image data exceeds a predetermined value. In this case, the imaging unit 120 and the image data acquisition unit 20 acquire image data more frequently than the frequency at which the position detection unit 30 generates position data.

[0053] Each sensor member 112 may have the same shape. The length of the pressing surface 116 of one sensor member 112 in the X-axis direction is defined as L, and the width in the Y-axis direction is defined as W. If the sensor members 112 are too small, there is a high possibility that one ground contact portion will span multiple sensor members 112. The pressing surface 116 of each sensor member 112 may be larger than the ground contact portion of each leg of the living body 200. The length L and width W of the pressing surface 116 of each sensor member 112 may be 20 cm or more. The length L and width W may also be 30 cm or more.

[0054] If the sensor element 112 becomes too large, the possibility of multiple legs stepping on one sensor element 112 at the same time increases. The sensor element 112 may not be able to separate the measured load value into components caused by multiple legs. The length L and width W of the pressing surface 116 may be smaller than the distance between the two legs of the living organism 200. This reduces the possibility of multiple legs stepping on one sensor element 112 at the same time. The length L and width W may be 50 cm or less. The length L may be larger than the width W. For example, if the living organism 200 is a cow, the distance between the front and rear legs is larger than the distance between the left and right legs. Therefore, even if the length L is large, the possibility of multiple legs stepping on one sensor element 112 at the same time is relatively small.

[0055] The sensor members 112 may be arranged in the X-axis direction over a range longer than the stride of the living body 200. The sensor members 112 may be arranged over a range of 2 m or more in the X-axis direction, or may be arranged over a range of 3 m or more.

[0056] The sensor member 112 may measure a static load (i.e., the instantaneous value of the load) or a dynamic load (i.e., the time waveform of the load). The sensor member 112 may measure a static load using a strain gauge or the like. The sensor member 112 may measure a dynamic load using a piezoelectric element or the like. The sensor member 112 may measure both static and dynamic loads. The sensor member 112 may be a plate-shaped sensor including at least one measuring element selected from the group consisting of an optical fiber sensor, a magnetic fiber sensor, a strain gauge, and a piezoelectric element.

[0057] The sensor member 112 may have a protective structure that protects the measuring element, such as a strain gauge or a piezoelectric element. The protective structure is, for example, a metal or rubber cover that covers the measuring element. By providing the protective structure, the measuring element can be protected from damage due to load or contamination by excrement.

[0058] Figure 4 shows an example of the results of measuring the load on each leg of the same individual. The horizontal axis of Figure 4 indicates the date on which the load was measured, and the vertical axis indicates the magnitude of the load. Figure 4 shows the magnitude of the load for each of the ground contact portions 211 to 214 corresponding to the four legs.

[0059] As described above, the state estimation unit 50 may estimate the state of the living body 200 based on a feature amount indicating the unevenness of the load on each leg. The unevenness of the load is indicated by a difference D1 between the reference value Lr and the amount of load on each leg of the living body (in the example of FIG. 4 , the left front leg, left hind leg, right front leg, and right hind leg). The reference value Lr may be the sum of the loads on each leg divided by the number of legs (or a value equal to ¼ of the body weight of the living body 200), or may be the time average value of the loads on the leg for which unevenness is to be calculated over a predetermined period of time. In another example, if the target leg is a front leg, the average value of the loads on the left front leg and the right front leg may be used as the reference value Lr. If the target leg is a hind leg, the average value of the loads on the left hind leg and the right hind leg may be used as the reference value Lr. In another example, the state estimation unit 50 may estimate that an abnormality has occurred in the leg in the ground-contact portion of the target leg when the difference D1 between the load measured for the target leg and the reference value Lr is equal to or greater than a predetermined reference value.

[0060] The state estimation unit 50 may estimate the state of the living organism 200 based on a feature indicating a difference from past measurement data of the same individual. The state estimation unit 50 may estimate that an abnormality has occurred in the leg being evaluated when a feature indicating a change in the current difference D1 from the past difference D1 is equal to or greater than a preset reference value. The state estimation unit 50 may estimate the state of the living organism 200 based on a time differential value of the change. For example, when the time differential value of the change is increasing, the state estimation unit 50 may determine that the living organism 200 is in a state requiring early intervention.

[0061] The state estimation unit 50 may compare the past feature vector of the same individual with the feature vector of the evaluation target using any distance measure (e.g., Euclidean distance). The feature vector may be obtained by directly compressing data using a machine learning model, or by compressing designed feature vectors using a machine learning model. It is also possible to use such vectors as training data to create and evaluate an anomaly detection model based on Mahalanobis distance or the like. In this case, a model is created using data from the living organism 200 in a normal state, and an appropriate threshold is set.

[0062] The state estimation unit 50 may estimate the state of the living organism 200 based on whether the walking tendency of the living organism 200 calculated from the measurement data and image data deviates from the past walking tendency. The walking tendency includes, for example, at least one of the stride length and walking speed of the living organism 200. When the difference between the current stride length or walking speed of the living organism 200 and the past stride length or walking speed is equal to or greater than a preset reference value, it may be estimated that an abnormality has occurred in the living organism 200.

[0063] Fig. 5 is a diagram showing an example of the load measurement results for multiple individuals. The horizontal axis of Fig. 5 indicates the date on which the load was measured, and the vertical axis indicates the magnitude of the load. Fig. 5 shows the load measurement results for organisms 200-1, 200-2, 200-3, and 200-4. The load measurement results for each organism 200 may be the load measurement results for a common leg.

[0064] The state estimation unit 50 may estimate the state of the living organism 200 based on a feature amount indicating a difference between the load measurement data of the individual to be determined (living organism 200-1 in the example of FIG. 5) and the load measurement data of other individuals (living organisms 200-2, 200-3, and 200-4 in the example of FIG. 5). The state estimation unit 50 may estimate the state of the living organism 200 based on a difference D2 between the measurement data of the individual to be determined and the average value of the measurement data of individuals other than the individual to be determined. The state estimation unit 50 may estimate that an abnormality has occurred in the individual to be determined when the difference D2 is equal to or greater than a predetermined reference value.

[0065] The condition estimation unit 50 may perform the following processes: (1) Collect multiple sets of load data from healthy individuals as reference data, and perform polynomial approximation on a graph of the load over time for each leg in the reference data. The load graph may be, for example, a graph like the one shown in FIG. 4 . (2) Obtain a graph of the load over time for each leg from the measurement data of the individual to be evaluated, and perform polynomial approximation on the graph for each leg. (3) Compare the coefficients of the polynomial related to the measurement data with the coefficients of the polynomial related to the reference data for each leg to estimate the condition of each leg of the individual to be evaluated. For example, the difference in coefficients for each degree of the polynomial may be calculated, and the magnitude of the difference may be used to determine whether or not an abnormality has occurred in the individual.

[0066] In the above-described process (3), the state estimation unit 50 may use principal component analysis to reduce the number of dimensions (i.e., the number of parameters) of the parameters to be compared and then calculate the distance between the parameters. For example, if the polynomial is tenth order and has ten coefficients, the state estimation unit 50 may use principal component analysis to generate two principal component parameters from the ten coefficients. The state estimation unit 50 may determine that an abnormality has occurred in the individual to be evaluated when the distance between two principal component parameters related to the measurement data and two principal component parameters related to the reference data is equal to or greater than a tolerance. Furthermore, the state estimation unit 50 may use a machine learning algorithm to compare previously learned normal parameters with the parameters to be compared, thereby determining an abnormality in the living organism 200. The machine learning algorithm may be, for example, one-class SVM, isolation forest, autoencoder, or principal component analysis.

[0067] In this example, the coefficients of the polynomial obtained by polynomial approximation of the change in load over time are used as the above-mentioned feature quantity. When performing polynomial approximation on the measurement data and the reference data, the approximation may be performed under the same conditions for the target data. For example, the state estimation unit 50 may perform polynomial approximation by aligning the scale of the time axis in the time-varying graph of the measurement data and the reference data. Furthermore, the state estimation unit 50 may perform polynomial approximation after normalizing the magnitude of the load or aligning the zero points of the load.

[0068] The state estimation unit 50 may estimate that an abnormality has occurred in the individual being assessed when the amount of change in the current difference D2 relative to the past difference D2 is equal to or greater than a preset reference value. The state estimation unit 50 may estimate the state of the individual being assessed based on a time differential value of the amount of change. For example, when the time differential value of the amount of change is increasing, the state estimation unit 50 may determine that the individual is in a state requiring early intervention.

[0069] Fig. 6 shows an example of a time waveform of the measurement result output by one sensor element 112. The sensor element 112 of this example is capable of measuring dynamic loads. Fig. 6 shows the time waveform for one step from when one of the legs of the living body 200 comes into contact with the pressing surface 116 of the sensor element 112 until it leaves the contact. The time waveform has two peaks P1 and P2 as an example, but the time waveform may have a single peak or more peaks.

[0070] The state estimation unit 50 may estimate the state of the living body 200 based on feature amounts extracted from a time waveform such as that shown in Fig. 6. The state estimation unit 50 may extract feature amounts corresponding to the foot contact time of the living body 200 from the time waveform. The foot contact time is the length of the period from when the foot comes into contact with the pressing surface 116 of the sensor member 112 until when the foot leaves the pressing surface 116.

[0071] The condition estimation unit 50 may extract, as a feature corresponding to the ground contact time, the time T1 from when the load reaches or exceeds a predetermined value until when it falls below the predetermined value. The predetermined value may be set according to a maximum value P1 of the load in the time waveform. As an example, the predetermined value may be α times the maximum value P1 (where α is greater than 0 and less than 1). As an example, α may be 0.1, or may be another value.

[0072] The condition estimation unit 50 may extract the time T2 from the first peak P1 to the last peak P2 in the time waveform as a feature corresponding to the ground contact time. In this case, the time waveform includes multiple peaks. The condition estimation unit 50 may estimate that an abnormality has occurred in the leg when the time T1 or the time T2 is equal to or less than a set reference value.

[0073] The state estimation unit 50 may estimate the state of the living body 200 based on at least one of the feature amounts described in Figures 4 to 6. However, the feature amounts for estimating the state of the living body 200 are not limited to the examples described in Figures 4 to 6.

[0074] FIG. 7 is a diagram showing another example of the load measurement results of one sensor member 112. In this example, each sensor member 112 measures the load distribution on the pressing surface 116. That is, the sensor member 112 measures the load distribution within one ground contact portion (ground contact portion 211 in FIG. 7 ). The spatial resolution of the load measurement on the pressing surface 116 may be 2 cm or less, or may be 1 cm or less, in both the X-axis direction and the Y-axis direction. In FIG. 7 , the magnitude of the load is schematically shown by the shade of the diagonal hatching. In the example of FIG. 7 , the magnitude of the load is shown in three values ​​(three shades of hatching), but the sensor member 112 may be capable of measuring the magnitude of the load in more than three values.

[0075] The state estimation unit 50 estimates the state of the living organism 200 based on the distribution of loads on the ground contact portions of each leg of the living organism 200. By measuring the distribution of loads within the ground contact portions, the state of the living organism 200 can be estimated in more detail.

[0076] The condition estimation unit 50 may estimate the type of hoof disease based on the distribution of load within the ground contact area. The type of hoof disease may be classified by the so-called disease name, such as sole ulcer, white band disease, or digital dermatitis, or may be classified by the location of the affected area, or may be classified by a user or other criteria.

[0077] When the living body 200 is affected by hoof disease or in the pre-hoof disease stage, the living body 200 tends to walk in a manner that reduces the load on the affected area. Therefore, the type of hoof disease can be estimated based on the distribution of load within the contact area. The condition estimation unit 50 may extract, as feature information, the magnitude of the load at a specific point within the contact area, the shape or size of an area within the contact area where the load is below a set value, the shape or size of an area within the contact area where the load is above a set value, the position of the center of gravity of the load within the contact area, the position within the contact area where the load is minimum, or changes in these parameters. The condition estimation unit 50 may estimate the type of hoof disease by comparing this feature information with preset reference information. The condition estimation unit 50 may estimate the type of hoof disease based on a combination of multiple feature information. The condition estimation unit 50 may be configured with a range of feature information for each type of hoof disease. If the feature information extracted from the measurement data belongs to any of the ranges of hoof disease, the condition estimation unit 50 may estimate that the living body 200 is in a state of the hoof disease.

[0078] The state estimation unit 50 may estimate the state of the living organism 200 based on a predetermined rule or algorithm. In another example, the state estimation unit 50 may estimate the state of the living organism 200 using a training model generated by machine learning. For example, the training model may be generated by machine learning using the load distribution on the ground contact portion of a living organism suffering from hoof disease as training data and the type of hoof disease of the living organism as label data. The state estimation unit 50 may input the load distribution of the living organism 200 to be estimated into the training model to estimate the type of hoof disease. Furthermore, the state estimation unit 50 may use a training model generated by machine learning using measurement data of the load on the ground contact portion of the living organism 200 as training data and the state of the living organism 200 as label data. In this case, the state estimation unit 50 estimates the state of the living organism 200 using the training model in the examples described with reference to FIGS. 1 to 6 . The generation of the training model will be described later.

[0079] FIG. 8 is a diagram showing an example of the position of the center of gravity of the load in the contact portion 211. FIG. 8 shows an initial position 221 of the center of gravity and a changed position 222. The initial position 221 is an example of the position of the center of gravity in a state without hoof disease. The changed position 222 is an example of the position of the center of gravity in a state with hoof disease. The condition estimation unit 50 may estimate the type of hoof disease using the change in the center of gravity position as characteristic information. The condition estimation unit 50 may estimate the type of hoof disease based on at least one of the direction from the initial position 221 toward the changed position 222 and the distance between the initial position 221 and the changed position 222.

[0080] As described in FIG. 8 , the condition estimation unit 50 may estimate the type of hoof disease suffered by an individual based on changes over time in characteristic information extracted from measurement data of the same individual. The condition estimation unit 50 may estimate the type of hoof disease suffered by the leg of the individual being estimated by comparing the characteristic information described in FIGS. 7 and 8 between the leg of the individual being estimated and other legs of the same individual. The comparison process between legs of the same individual is similar to the example in FIG. 4 . The condition estimation unit 50 may estimate the type of hoof disease suffered by the individual being estimated by comparing the characteristic information described in FIGS. 7 and 8 between the individual being estimated and multiple other individuals. The comparison process between individuals is similar to the example in FIG. 5 .

[0081] In the example described in Figures 7 and 8, the condition estimation unit 50 may estimate the type of hoof disease of the living organism 200 without using image data of the living organism 200. In this case, the analysis device 100 may not include the imaging unit 120, the image data acquisition unit 20, and the position detection unit 30. In another example, even in the example described in Figures 7 and 8, the condition estimation unit 50 may estimate the type of hoof disease of the living organism 200 using image data of the living organism 200. Based on the image data, the condition estimation unit 50 may determine which leg of the living organism 200 each contacts with the ground extracted from the measurement data. This makes it easier to determine which leg of the living organism 200 is affected by hoof disease.

[0082] The condition estimation unit 50 may use the load distribution on the ground contact portion to determine a condition of the living organism 200 other than the type of hoof disease. The condition estimation unit 50 may determine a condition such as lameness of the living organism 200 based on the load distribution.

[0083] Fig. 9 is a diagram showing another example of the configuration of analysis device 100. Analysis device 100 of this example further includes a speed detection unit 40 in addition to the configuration of analysis device 100 shown in Fig. 1. The other configuration of analysis device 100 is similar to that of any of the aspects of analysis device 100 described with reference to Figs. 1 to 8.

[0084] The speed detection unit 40 detects the moving speed of the living body 200 based on the image data. The image data in this example includes images showing changes in the position of the living body 200 at multiple times. The state estimation unit 50 in this example estimates the state of the living body 200 based further on the moving speed of the living body 200.

[0085] The state estimation unit 50 may estimate the state of the living organism 200 by comparing the movement speed of the living organism 200 with a reference value. For example, the state estimation unit 50 may use the past movement speed of the same individual or the movement speed of another individual as the reference value. When the movement speed of the living organism 200 becomes slower than the past movement speeds or the movement speeds of other individuals, the state estimation unit 50 may estimate that there is a high possibility that some abnormality has occurred in the living organism 200.

[0086] In another example, the state estimation unit 50 may select measurement data to be used to estimate the state of the living organism 200 using the moving speed of the living organism 200. Each measurement data may be stored in association with the moving speed of the living organism 200 at the time of measurement. For example, even for measurement data of the same individual, the feature quantities and feature information described in FIGS. 1 to 8 may change between a stationary state and a moving state. When comparing past and current measurement data of the same individual, if the current measurement data is obtained by measuring the living organism 200 in a stationary state, the state estimation unit 50 may extract past measurement data obtained by measuring the living organism 200 in a stationary state. Similarly, if the current measurement data is obtained by measuring the living organism 200 in a moving state, the state estimation unit 50 may extract past measurement data obtained by measuring the living organism 200 in a moving state. Even when comparing measurement data between the individual to be estimated and another individual, the state estimation unit 50 may extract measurement data that share a common stationary or moving state.

[0087] The state estimation unit 50 may select measurement data used to estimate the state of the living organism 200 according to the magnitude of the moving speed. For example, even in measurement data of the same individual, the feature amounts and feature information described with reference to Figures 1 to 8 may change between a state in which the individual is moving fast and a state in which the individual is moving slowly. The state estimation unit 50 may extract and compare measurement data with similar magnitudes of moving speed.

[0088] The speed detection unit 40 may detect, based on the image data, the acceleration of the living body 200. In the example described in Fig. 9, the state estimation unit 50 may use the acceleration of the living body 200 instead of the moving speed of the living body 200.

[0089] FIG. 10 is a diagram showing another example of the configuration of the analysis device 100. The analysis device 100 of this example further includes a weight calculation unit 60 in addition to the configuration of the analysis device 100 of any of the aspects described in FIGS. 1 to 9. The weight calculation unit 60 calculates the weight of the living organism 200 based on the measurement data. The weight calculation unit 60 may calculate the weight of the living organism 200 from the total load on each leg of the living organism 200 in a stationary state. The weight calculation unit 60 may notify the administrator of the living organism 200, etc., of the calculated weight. The weight calculation unit 60 may notify the state estimation unit 50 of the calculated weight.

[0090] The state estimation unit 50 may estimate the state of the living organism 200 based on the weight of the living organism 200. The state estimation unit 50 may estimate the state of the living organism 200 based on a change in the weight of the living organism 200. The state estimation unit 50 may estimate that an abnormality has occurred in the living organism 200 when a change in the weight of the living organism 200 within a predetermined period exceeds a set reference value.

[0091] In another example, the state estimation unit 50 may use the weight of the living body 200 to select measurement data to be used to estimate the state of the living body 200. Each measurement data may be stored in association with the weight of the living body 200 at the time of measurement. For example, even if the measurement data is of the same individual, the feature amount (e.g., the magnitude of the load) and feature information described in FIGS. 1 to 8 may change depending on changes in weight. When comparing past and current measurement data of the same individual, the state estimation unit 50 may compare measurement data with similar weights. Similarly, when comparing measurement data between an individual to be estimated and another individual, the state estimation unit 50 may compare measurement data with similar weights.

[0092] Fig. 11 is a block diagram showing an example of a training device 300 that trains a training target model. When measurement data including a distribution of loads on the living organism 200 is input, the training target model outputs the state of the living organism 200. The state estimation unit 50 described with reference to Figs. 1 to 10 may estimate the state of the living organism 200 using the training target model.

[0093] The training device 300 trains a training target model through machine learning using training data. The training data may include a distribution of loads on the ground contact portions of each leg of the living body 200. The training data may include at least a portion of the measurement data and image data described in FIGS. 1 to 10, or may include data generated from the measurement data and image data. The data generated from the measurement data and image data may be the above-mentioned feature amount or feature information, or may be the moving speed or weight of the living body 200.

[0094] The training device 300 of this example includes a training data acquisition unit 310, a data processing unit 340, an evaluation unit 320, and a training unit 330. At least a part of the training data acquisition unit 310, the evaluation unit 320, and the training unit 330 may be a computer. A program for causing the computer to function as the training device 300 may be installed in the computer. The program may be stored in a computer-readable medium such as a memory, or may be transmitted to the computer via a communication line.

[0095] The training data acquisition unit 310 acquires the training data and the identification information of each individual in association with each other. The training data may be generated from measurement data of the actual living organism 200.

[0096] The data processing unit 340 performs a predetermined process on the training data. The data processing unit 340 estimates which leg of the living organism 200 the load distribution included in the training data corresponds to, and includes the estimation result in the training data. The data processing unit 340 may estimate which leg of the living organism 200 the data included in the training data corresponds to, based on image data of the living organism 200. The data processing unit 340 may estimate which leg of the living organism 200 each contact portion corresponds to, based on the relative position of each contact portion included in the training data.

[0097] The data processing unit 340 may further perform at least one of noise removal and upsampling or downsampling of a data set in a spatial or temporal series on the training data. Noise removal may be a process of removing a predetermined frequency component from a data sequence included in the training data. Upsampling may be a process of interpolating data on a spatial or temporal axis from a data sequence included in the training data. Downsampling may be a process of thinning data on a spatial or temporal axis from a data sequence included in the training data.

[0098] 1 to 10 from the measurement data and image data included in the training data, the data processing unit 340 may extract at least one of the feature amounts, feature information, movement speed, and weight described in Figures 1 to 10 and include it in the training data. By using feature amounts, etc. as training data, it may be possible to create a model with less learning data than when using measured load values ​​directly as training data (reducing the amount of calculations and reducing the memory capacity required to store the model to be trained), or it may be easier to interpret the inference results of the model.

[0099] The evaluation unit 320 inputs the training data into the training target model. The evaluation unit 320 may input the training data including the above-mentioned estimation result by the data processing unit 340 into the training target model. The training target model estimates and outputs the condition of the living organism 200 according to the input training data. The condition of the living organism 200 may be the presence or absence of an abnormality such as lameness, or may be the type of hoof disease. The evaluation unit 320 obtains an output result indicating the condition of the living organism 200 from the training target model.

[0100] The training unit 330 adjusts the training target model based on the label data indicating the state of the individual corresponding to the identification information and the output result. The training unit 330 adjusts the training model so that the output result of the training target model approaches the label data. The adjustment of the training target model may be a process of adjusting parameters such as weighting coefficients in the calculation processing of the training target model.

[0101] The label data may be generated from the results of observing the state of the actual living organism 200. The label data may include the results of observing the state of the living organism 200 after the time point of measurement of the measurement data included in the training data. This makes it possible to generate a training object model that can estimate a precursor to an abnormality such as lameness. The label data may include the results of observing the state of the living organism 200 at different points in time. In other words, the label data may include results that indicate the time series transition of the state of the living organism 200. The training object model may estimate the time series transition of the state of the living organism 200 when measurement data is input.

[0102] The training device 300 may generate a training object model using training data and label data of a living organism 200 in which an abnormality is detected. The training device 300 may also generate a training object model using training data of a living organism 200 in which no abnormality is detected. In this case, label data may not be used. In this case, the training object model estimates the occurrence of an abnormality when measurement data input as an estimation object deviates from the measurement data included in the training data. The training device 300 may generate a training object model using training data and label data of a living organism 200 in which an abnormality is detected and a living organism 200 in which no abnormality is detected.

[0103] The training model may output a score indicating the degree of abnormality such as lameness, etc. If the score exceeds a threshold, analysis device 100 may notify the user of the occurrence of the abnormality.

[0104] The label data may include information indicating the cause of an abnormality such as lameness or hoof disease. The information may be generated based on a diagnosis by a veterinarian or the like. When an abnormality is predicted in the living body 200, the training model may also predict the cause of the abnormality.

[0105] Various embodiments of the present invention may be described with reference to flowcharts and block diagrams, where the blocks may represent (1) stages of a process in which operations are performed or (2) sections of apparatus responsible for performing the operations. Particular stages and sections may be implemented by dedicated circuitry, programmable circuitry provided with computer-readable instructions stored on a computer-readable medium, and / or a processor provided with computer-readable instructions stored on a computer-readable medium. Dedicated circuitry may include digital and / or analog hardware circuitry, and may include integrated circuits (ICs) and / or discrete circuits. Programmable circuitry may include reconfigurable hardware circuitry including logical AND, OR, XOR, NAND, NOR, and other logic operations, flip-flops, registers, memory elements such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), and the like.

[0106] A computer-readable medium may include any tangible device capable of storing instructions that are executed by an appropriate device, such that the computer-readable medium having instructions stored thereon comprises an article of manufacture containing instructions that can be executed to create means for performing the operations specified in the flowcharts or block diagrams. Examples of computer-readable media may include electronic, magnetic, optical, electromagnetic, and semiconductor storage media. More specific examples of computer-readable media may include floppy disks, diskettes, hard disks, random access memories (RAMs), read-only memories (ROMs), erasable programmable read-only memories (EPROMs or flash memories), electrically erasable programmable read-only memories (EEPROMs), static random access memories (SRAMs), compact disc read-only memories (CD-ROMs), digital versatile discs (DVDs), Blu-ray (RTM) discs, memory sticks, integrated circuit cards, and the like.

[0107] The computer readable instructions may include either assembler instructions, Instruction Set Architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state-setting data, or source or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk®, JAVA®, C++, etc., and conventional procedural programming languages ​​such as the “C” programming language or similar programming languages.

[0108] The computer-readable instructions may be provided to a processor or programmable circuit of a programmable data processing device, such as a general-purpose computer, a special-purpose computer, or another computer, either locally or via a wide-area network (WAN) such as a local area network (LAN) or the Internet, which executes the computer-readable instructions to create means for performing the operations specified in the flowcharts or block diagrams. Here, the computer may be a personal computer (PC), a tablet computer, a smartphone, a workstation, a server computer, a general-purpose computer, a special-purpose computer, or the like, or may be a computer system in which multiple computers are connected. Such a computer system in which multiple computers are connected is also called a distributed computing system, and is a broad definition of computer. In a distributed computing system, multiple computers collectively execute a program by each executing a portion of the program and passing data between the computers as needed during program execution.

[0109] Examples of processors include computer processors, central processing units (CPUs), processing units, microprocessors, digital signal processors, controllers, microcontrollers, etc. A computer may have one processor or multiple processors. In a multiprocessor system with multiple processors, each processor executes a portion of a program and passes data between processors as needed during program execution, allowing the multiple processors to collectively execute the program. For example, in multitasking, each of the multiple processors may execute a portion of each task in small chunks by switching tasks at each time slice. In this case, which portion of a program each processor executes changes dynamically. Which portion of a program each of the multiple processors executes may also be statically determined by multiprocessor-aware programming.

[0110] 12 illustrates an example of a computer 1200 in which aspects of the present invention may be embodied, in whole or in part. Programs installed on the computer 1200 may cause the computer 1200 to function as or perform operations associated with an apparatus or one or more sections of the apparatus according to embodiments of the present invention, and / or to perform a process or steps of a process according to embodiments of the present invention. Such programs may be executed by the CPU 1212 to cause the computer 1200 to perform specific operations associated with some or all of the blocks in the flowcharts and block diagrams described herein.

[0111] A computer 1200 according to this embodiment includes a CPU 1212, a RAM 1214, a graphics controller 1216, and a display device 1218, which are interconnected by a host controller 1210. The computer 1200 also includes input / output units such as a communication interface 1222, a storage device 1224 such as a hard disk drive, a DVD-ROM drive 1226, and an IC card drive, which are connected to the host controller 1210 via an input / output controller 1220. The computer also includes legacy input / output units such as a ROM 1230 and a keyboard 1242, which are connected to the input / output controller 1220 via an input / output chip 1240.

[0112] The CPU 1212 operates according to programs stored in the ROM 1230 and the RAM 1214, thereby controlling each unit. The graphics controller 1216 acquires image data generated by the CPU 1212 into a frame buffer or the like provided in the RAM 1214 or into the graphics controller 1216 itself, and causes the image data to be displayed on the display device 1218.

[0113] The communication interface 1222 communicates with other electronic devices via a network. The storage device 1224 stores programs and data used by the CPU 1212 in the computer 1200. The DVD-ROM drive 1226 reads programs or data from a DVD-ROM 1227 and provides the programs or data to the storage device 1224 via the RAM 1214. The IC card drive reads programs and data from an IC card and / or writes programs and data to an IC card.

[0114] The ROM 1230 stores therein a boot program or the like that is executed by the computer 1200 upon activation, and / or programs that depend on the hardware of the computer 1200. The input / output chip 1240 may also connect various input / output units to the input / output controller 1220 via a parallel port, a serial port, a keyboard port, a mouse port, etc.

[0115] The programs are provided by a computer-readable medium such as a DVD-ROM 1227 or an IC card. The programs are read from the computer-readable medium, installed in the storage device 1224, RAM 1214, or ROM 1230, which are also examples of computer-readable media, and executed by the CPU 1212. Information processing described in these programs is read by the computer 1200, and brings about cooperation between the programs and the various types of hardware resources described above. An apparatus or a method may be configured by implementing information manipulation or processing in accordance with the use of the computer 1200.

[0116] For example, when communication is performed between computer 1200 and an external device, CPU 1212 may execute a communication program loaded into RAM 1214 and instruct communication interface 1222 to perform communication processing based on the processing described in the communication program. Under the control of CPU 1212, communication interface 1222 reads transmission data stored in a transmission buffer processing area provided in RAM 1214, storage device 1224, DVD-ROM 1227, or a recording medium such as an IC card, and transmits the read transmission data to the network, or writes received data received from the network to a reception buffer processing area or the like provided on the recording medium.

[0117] The CPU 1212 may read all or a necessary portion of a file or database stored in an external recording medium such as the storage device 1224, the DVD-ROM drive 1226 (DVD-ROM 1227), an IC card, etc. into the RAM 1214, and perform various types of processing on the data on the RAM 1214. The CPU 1212 then writes back the processed data to the external recording medium.

[0118] Various types of information, such as various types of programs, data, tables, and databases, may be stored on the recording medium and may undergo information processing. The CPU 1212 may perform various types of processing on data read from the RAM 1214, including various types of operations, information processing, conditional judgment, conditional branching, unconditional branching, information search / replacement, etc., as described throughout this disclosure and specified by the instruction sequences of the programs, and write the results back to the RAM 1214. The CPU 1212 may also search for information in a file, database, etc. on the recording medium. For example, if multiple entries each having an attribute value of a first attribute associated with an attribute value of a second attribute are stored on the recording medium, the CPU 1212 may search for an entry that matches a condition specified by the attribute value of the first attribute from among the multiple entries, read the attribute value of the second attribute stored in the entry, and thereby obtain the attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

[0119] The above-described programs or software modules may be stored in a computer-readable medium on or near the computer 1200. A recording medium such as a hard disk or RAM provided in a server system connected to a dedicated communication network or the Internet can also be used as a computer-readable medium, thereby providing the programs to the computer 1200 via the network.

[0120] Although the present invention has been described above using embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments. It will be apparent to those skilled in the art that various modifications and improvements can be made to the above embodiments. It is clear from the claims that such modifications and improvements can also be included within the technical scope of the present invention.

[0121] It should be noted that the order of execution of each process, such as operations, procedures, steps, and stages, in the devices, systems, programs, and methods shown in the claims, specifications, and drawings is not specifically stated as "before," "prior to," etc., and that the processes can be performed in any order unless the output of a previous process is used in a subsequent process. Even if the operational flow in the claims, specifications, and drawings is described using "first," "next," etc. for convenience, this does not mean that the processes must be performed in this order.

[0122] 10...Load data acquisition unit, 20...Image data acquisition unit, 30...Position detection unit, 40...Speed ​​detection unit, 50...State estimation unit, 60...Weight calculation unit, 100...Analysis device, 110...Sensor unit, 112...Sensor member, 114...Path, 116...Pressing surface, 120...Image capture unit, 200...Body, 202...Tag, 211, 212, 213, 214...Ground contact portion, 221...Initial position, 222...Changed position, 300...Training device, 310...Training data acquisition unit, 320...Evaluation unit, 330...Training unit, 340...Data processing unit

Claims

1. An analysis device comprising: a load data acquisition unit that acquires measurement data from a plurality of sensor elements that measure load and are arranged on a path along which a living organism walks, in association with identification information for each individual living organism; an image data acquisition unit that acquires images of the living organism along the path together with the identification information; a position detection unit that generates, based on the images, position data indicating which of the plurality of sensor elements is pressed by each leg of the living organism; and a state estimation unit that estimates the state of the living organism based on the measurement data and the position data.

2. The analysis device described in claim 1, wherein the position detection unit determines whether or not the ground contact portion of any leg of the living body spans two or more of the sensor elements, and the state estimation unit, when the ground contact portion spans two or more of the sensor elements, calculates the measurement value of the load of the ground contact portion based on the measurement value of the load at the two or more sensor elements, and when the ground contact portion does not span two or more of the sensor elements, calculates the measurement value of the load of the ground contact portion based on the measurement value of the load at one of the sensor elements.

3. The analytical device according to claim 1, wherein each of the plurality of sensor elements is in the shape of a plate having a pressing surface.

4. The analysis device according to claim 1, wherein each of the plurality of sensor elements is plate-shaped with a pressure surface, each of the sensor elements measures the distribution of load on the pressure surface, and the state estimation unit estimates the state of the living organism based on the distribution of load on the ground contact portion of each leg of the living organism.

5. The analysis device according to claim 1, further comprising a speed detection unit that detects the moving speed of the living body based on the image, and the state estimation unit estimates the state of the living body further based on the moving speed of the living body.

6. An analysis device comprising: a load data acquisition unit that is arranged on a path along which a living organism walks and acquires measurement data from a sensor element that measures the distribution of load on the ground contact portions of each leg of the living organism, in association with identification information for each individual living organism; and a state estimation unit that estimates the state of the living organism based on the distribution of load on the ground contact portions of each leg of the living organism.

7. The analysis device according to claim 6, wherein the condition estimation unit estimates the type of hoof disease of the living organism.

8. The analysis device according to claim 1 or 6, wherein the state estimation unit estimates the state of the living organism based on at least one feature of uneven load on each leg, differences from past measurement data of the same individual, and differences from measurement data of other individuals.

9. The analysis device according to claim 1 or 6, further comprising a weight calculation unit that calculates the weight of the living body based on the measurement data.

10. The analysis device according to claim 1 or 6, wherein the state estimation unit selects an estimation method for estimating the state of the living organism based on the estimation result of whether the living organism is walking or stationary.

11. A training device comprising: a training data acquisition unit that acquires training data including the distribution of loads on the ground contact portions of each leg of a living organism in association with identification information for each individual of the living organism; an evaluation unit that inputs the training data into a training target model and acquires output results indicating the state of the living organism from the training target model; and a training unit that adjusts the training target model based on label data indicating the state of the individual corresponding to the identification information and the output results.

12. The training device according to claim 11, wherein the training subject model outputs the type of hoof disease of the living organism as the state of the living organism.

13. A training device as described in claim 11 or 12, further comprising a data processing unit that estimates which leg of the living body the load distribution included in the training data corresponds to and includes the estimation result in the training data, and the evaluation unit inputs the training data including the estimation result into the training target model.

14. A training device as described in claim 11 or 12, further comprising a data processing unit that estimates which leg of the living body the load distribution included in the training data corresponds to and includes the estimation result in the training data, and the evaluation unit inputs the training data including the estimation result into the training target model.

15. A program for causing a computer to function as the analysis device according to claim 1 or 6.

Images