Beef cattle body condition information measuring system and device

Abstract

The invention discloses a beef cattle body condition information measuring system and device, and relates to the technical field of animal husbandry management, and the system comprises a body condition measuring module which is used for comparing the collected overlooking cattle back images of the nth week and the (n + 1) th week to obtain cattle back sign change data; the health classification module is used for collecting beef cattle diet data from the nth week to the (n + 1) th week and conducting pathological judgment on the beef cattle diet data and the cattle back sign change data based on a preset classification model so as to obtain health labels of the beef cattle, and the health labels comprise healthy beef cattle, sub-healthy beef cattle and sick beef cattle; the disease prediction module is used for obtaining environment data from the nth week to the (n + 1) th week, inputting the diet data and the environment data of the individuals with the health labels as sub-health beef cattle into a preset regression model to obtain the disease probability of the beef cattle from the (n + 1) th week to the (n + 2) th week, according to the invention, monitoring of beef cattle body conditions, accurate identification of health states and risk early warning of key individuals can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of animal husbandry, and particularly relates to a beef cattle body condition information measurement system and device. Background Art

[0002] With the increasing trend of the intelligent and large-scale development of animal husbandry, how to achieve refined management and real-time monitoring of the health status of individual beef cattle has become one of the key issues in breeding management. Traditional methods for judging the health of beef cattle mostly rely on manual observation, regular weighing, and feeding records, which have problems such as data lag, large subjective judgment errors, and difficulty in achieving early warning, and cannot effectively meet the management requirements of modern farms that emphasize both efficiency and benefit.

[0003] Currently, some breeding enterprises have tried to introduce image recognition, sensor detection, and Internet of Things technologies to build an individual-oriented health monitoring system. For example, existing solutions have tried to install camera devices to monitor beef cattle through video, and use image algorithms to identify abnormal behavior characteristics; or conduct status assessment based on behavior data such as water intake and feed intake. However, such solutions still have the following deficiencies: Most existing image recognition systems only conduct preliminary recognition of the posture or activity status of beef cattle, lacking dynamic extraction and analysis of the body surface growth characteristics, and cannot accurately reflect the trend of body condition changes; For the health assessment of beef cattle by combining body condition information, most use a single data index for judgment. For example, only based on abnormal food intake for status assessment, it is easily interfered by environmental and feed changes, and the false alarm rate is relatively high; Lack of accurate recognition ability for the "sub-healthy" stage and a prediction mechanism for subsequent disease risks, making it difficult to conduct early intervention on key individuals. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides a beef cattle body condition information measurement system and device.

[0005] Among them, a beef cattle body condition information measurement system includes: A body condition measurement module: used to compare the top-down view of the beef cattle's back collected in the nth week and the (n + 1)th week to obtain the data on the changes in the back signs of the cattle; A health classification module, used to collect the beef cattle's diet data from the nth week to the (n + 1)th week, and conduct pathological determination on the beef cattle's diet data and the data on the changes in the back signs of the cattle based on a preset classification model to obtain the health label of the beef cattle, and the health label includes "healthy beef cattle", "sub-healthy beef cattle", and "diseased beef cattle"; An incidence prediction module, configured to obtain environmental data within the nth week to the (n + 1)th week, and input the dietary data and environmental data of individuals with a health label of "sub-healthy beef cattle" into a preset regression model to obtain the incidence probability of beef cattle within the (n + 1)th week to the (n + 2)th week.

[0006] Further, the acquisition logic of the cattle back physical sign change data is as follows: Identify the key points of the cattle ridge midline and the key points of the cattle tail part in the top-down view of the cattle back, and extract the edge point set M of the body outline in the top-down view of the cattle back. The edge point set M includes m edge points, where m is an integer greater than 0; Connect the key points of the cattle ridge midline to obtain the cattle ridge curve, respectively draw the nearest perpendicular lines from the m edge points to the cattle ridge curve, select the longest nearest perpendicular line, and denote the corresponding edge point as P1. Establish a multi-segment plane coordinate system with the cattle ridge curve as the y-axis, and obtain the edge point P2 through edge point screening; Connect the key points of the cattle tail part to obtain the cattle tail curve, use the intersection point of the cattle tail curve and the cattle ridge curve as the connection point P3, connect the line segments in the order of P1, P3, and P2, and do not connect between P1 and P2 to obtain ∠α; Take ∠α in the nth week n and ∠α in the (n + 1)th week n+1 Perform a difference calculation to obtain ∠α cy , ∠α cy =∠α n+1 -∠α n , and take ∠α cy as the cattle back physical sign change data for output.

[0007] Further, the logic of the edge point screening is as follows: Arrange the m nearest perpendicular lines in descending order of length, and select the first two items and mark them as the longest perpendicular line i(1) and the second longest perpendicular line i(u), where u ≥ 2 and u is an integer; Establish a multi-segment plane coordinate system with the cattle ridge curve as the y-axis, denote the coordinates of the edge point P1 corresponding to the longest perpendicular line i(1) as (X1, Y1), and denote the coordinates of the edge point corresponding to the second longest perpendicular line i(u) as (Xu, Yu); Compare X1 with Xu. If X1 < 0 and Xu < 0, or X1 > 0 and Xu > 0, it is determined that the longest perpendicular line i(1) and the second longest perpendicular line i(u) are on the same side, and go to step S124; If X1 < 0 and Xu < 0, or X1 > 0 and Xu > 0, it is determined that the longest i(1) and the second longest perpendicular line i(u) are on different sides, and output the edge point of the second longest perpendicular line i(u) as P2; Select the nearest perpendicular line whose length is second only to the second-longest perpendicular line as the alternative perpendicular line, denoted as i(u + 1), let the second-longest perpendicular line i(u)= the alternative perpendicular line i(u + 1), and go to step S122.

[0008] Further, the beef cattle diet data includes the average daily food intake, the average daily water intake, and the average daily rumination frequency.

[0009] Further, the construction logic of the preset classification model is as follows: Obtain historical beef cattle classification data, and divide the historical beef cattle classification data into a training set and a test set. The historical beef cattle classification data includes beef cattle diet data, cattle back physical sign change data, and their corresponding health labels; Configure an initial classifier, use the beef cattle diet data and the cattle back physical sign change data in the training set as the input data of the initial classifier, and use the corresponding health labels in the training set as the output data of the initial classifier to train the initial classifier to obtain an initial classification network; Verify the initial classification network through the test set, and output the initial classification network whose accuracy is greater than or equal to the preset test accuracy as the pre-constructed classification model.

[0010] Further, the generation logic of the health label is as follows: Compare the average daily food intake, the average daily water intake, the average daily rumination frequency of the beef cattle, and the cattle back physical sign change data ∠α cy with their corresponding preset normal ranges. If the average daily food intake, the average daily water intake, the average daily rumination frequency, and the cattle back physical sign change data ∠α cy are all within the corresponding preset normal ranges, then mark the corresponding beef cattle as "healthy beef cattle"; If one of the average daily food intake, the average daily water intake, the average daily rumination frequency, and the cattle back physical sign change data ∠α cy is not within the preset normal range, then mark it as "sub-healthy beef cattle"; If two or more of the average daily food intake, the average daily water intake, the average daily rumination frequency, and the cattle back physical sign change data ∠α cy are not within the preset normal ranges, then mark it as "diseased beef cattle".

[0011] Further, the environmental data includes a temperature change trend graph, a humidity change trend graph, and a wind speed change trend graph of the breeding farm.

[0012] Further, the construction logic of the regression model is as follows: Obtain historical regression data, and divide the historical regression data into a test set and a training set. The historical regression data includes environmental data, diet data, and their corresponding disease incidence probabilities; Construct a regression network, using the environmental data and diet data in the training set as the input of the regression network, and using the corresponding incidence probability in the training set as the output data of the regression network to obtain an initial regression network; Validate the initial regression network through a test set, and output the initial regression network with a test error less than or equal to the preset test error as the regression model.

[0013] Furthermore, the acquisition logic of the incidence probability is as follows: Obtain historical data for calculating the incidence probability, which includes the value of weight change, ammonia concentration in the cowshed, and moisture content of dry food; Compare the value of weight change, ammonia concentration in the cowshed, and moisture content of dry food with the corresponding normal thresholds, and generate an incidence probability based on the corresponding set weight factors. The calculation formula for the incidence probability is: ι = ×γ1 + ×γ2 + ×γ3; Where ι is the incidence probability, γ1 + γ2 + γ3 = 1.24, Wc is the value of weight change, Nc is the ammonia concentration in the cowshed, and Md is the moisture content of dry food; Wy is the normal threshold of the value of weight change, Ny is the normal threshold of the ammonia concentration in the cowshed, and My is the normal threshold of the moisture content of dry food; γ1, γ2, and γ3 are the weight factors of the value of weight change, ammonia concentration, and moisture content of dry food respectively, and γ1 + γ2 + γ3 = 1.24.

[0014] A beef cattle body condition information measurement device for implementing a beef cattle body condition information measurement system according to any one of the above, characterized in that it includes the following components: An image acquisition component, which is arranged at the top of the channel for detecting the body condition of beef cattle, and is used to acquire a top-down view of the beef cattle's back; A diet data acquisition component, which includes a feeding detection sensor arranged at the bottom of the food basin, a water intake detection sensor arranged at the bottom of the water trough, and a frequency sensor arranged on the cow's neck, and is used to acquire the feeding amount, water intake amount, and rumination frequency of individual beef cattle; An environmental data acquisition component, which includes a temperature sensor, a humidity sensor, and a wind speed sensor arranged in the cowshed, and is used to acquire the temperature, humidity, and wind speed in the cowshed; A trend chart generation unit, which is used to read the collected temperature, humidity, and wind speed, and generate corresponding temperature change trend charts, humidity change trend charts, and wind speed change trend charts; A central processing unit, which calls a classification model and a regression model to perform data analysis on the collected various data, and outputs a health label and an incidence probability to a display unit; A storage unit for storing various types of collected data and responding to a trend graph generation unit and a central processing unit; A display unit for receiving the health label and the morbidity probability output by the central processing unit in real time and displaying them on the mobile terminal.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: By collecting the top-down images of the cow's back for two consecutive weeks to extract the data on the changes in the cow's back signs, and combining the average daily food intake, daily water intake, and daily rumination frequency of the beef cattle during the same period, a classification model is established to determine the health status of the beef cattle, realizing the accurate differentiation of individuals of "healthy beef cattle", "sub-healthy beef cattle", and "diseased beef cattle", and then managing the identified individuals in an abnormal state, achieving the determination of the physiological health status of beef cattle during growth when measuring the body condition of beef; For the beef cattle identified as "sub-healthy", further collect the data on the changes in temperature, humidity, and wind speed during the corresponding period, and input the diet records into the regression model to calculate the morbidity probability for the next week; thus, realizing the early prediction of the morbidity risk of key individuals, facilitating targeted early intervention and treatment in breeding management, and avoiding the occurrence of a large number of sick beef cattle; In summary, the present invention combines the top-down cow back image recognition and diet behavior data analysis, establishes a health status classification model, and further combines the environmental change trend and the regression prediction model to realize the assessment of the morbidity probability of sub-healthy individuals, and can realize the monitoring of the body condition of beef cattle, the accurate identification of the health status, and the risk warning of key individuals, improving the intelligent level of breeding management and the preposition of intervention decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings.

[0017] Figure 1 It is a module diagram of a beef cattle body condition information measurement system provided in Embodiment 1 of the present invention; Figure 2 It is a top-down image of the cow's back of a beef cattle body condition information measurement system provided in Embodiment 1 of the present invention Figure 3 It is a component diagram of a beef cattle body condition information measurement device provided in Embodiment 2 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0019] Embodiment 1 Please refer to Figure 1 As shown in the figure, this embodiment publicly provides a beef cattle body condition information measurement system, and the system includes: Body condition measurement module S100: used to compare the top-down view of the cattle's back collected in the nth week and the (n + 1)th week to obtain the change data of the cattle's back signs; It should be noted that: the top-down view of the cattle's back is as Figure 2 shown. The top-down view of the cattle's back is obtained through an image acquisition device, and the image acquisition device includes but is not limited to a fixed high-definition camera and a 3D depth camera. Among them, the acquisition time of the top-down view of the cattle's back is before morning feeding, and the acquisition cycle from the acquisition date of the new week to the acquisition date of the previous week is seven days; Specifically, the acquisition logic of the change data of the cattle's back signs is as follows: S110, identify the key points of the cattle's spine midline and the key points of the cattle's tail part in the top-down view of the cattle's back, and extract the edge point set M of the body outline in the top-down view of the cattle's back. The edge point set M includes m edge points, and m is an integer greater than 0; It should be noted that: the key points of the cattle's spine midline and the key points of the cattle's tail part can be identified through DeepLabCut, and the edge point set of the body outline of the beef cattle can be extracted through an edge detection algorithm; S120, connect the key points of the cattle's spine midline to obtain the cattle's spine curve, respectively draw the nearest perpendicular lines from the m edge points to the cattle's spine curve, select the longest nearest perpendicular line, and record the corresponding edge point as P1. Establish a multi-segment plane coordinate system with the cattle's spine curve as the y-axis, and obtain the edge point P2 through edge point screening; It should be noted that: the cattle's spine curve is obtained through curve fitting. Since there are inflection points in the cattle's spine curve, a multi-segment plane coordinate system needs to be established to accurately distinguish the left and right sides of the beef cattle's back; Specifically, the logic of the edge point screening is as follows: S121, arrange the m nearest perpendicular lines in descending order of length, select the first two items and mark them as the longest perpendicular line i(1) and the second longest perpendicular line i(u), where u ≥ 2 and u is an integer; S122. Establish a multi-segment plane coordinate system with the bovine spine curve as the y-axis. Denote the coordinates of the edge point P1 corresponding to the longest perpendicular line i(1) as (X1, Y1), and denote the coordinates of the edge point corresponding to the second-longest perpendicular line i(u) as (Xu, Yu). S123. Compare X1 with Xu. If X1 < 0 and Xu < 0, or X1 > 0 and Xu > 0, then determine that the longest perpendicular line i(1) and the second-longest perpendicular line i(u) are on the same side, and go to step S124. If X1 < 0 and Xu < 0, or X1 > 0 and Xu > 0, then determine that the longest i(1) and the second-longest perpendicular line i(u) are on different sides, and output the edge point of the second-longest perpendicular line i(u) as P2. S124. Select the nearest perpendicular line whose length ranking is just after the second-longest perpendicular line as the alternative perpendicular line, mark it as i(u + 1), let the second-longest perpendicular line i(u) = the alternative perpendicular line i(u + 1), and go to step S122. S130. Connect the key points at the bovine tail part to obtain the bovine tail curve. Take the intersection point of the bovine tail curve and the bovine spine curve as the connection point P3, and connect the line segments in the order of P1, P3, P2. Do not connect between P1 and P2 to obtain ∠α (as Figure 2 shown). S140. Calculate the difference between ∠α of the nth week n and ∠α of the (n + 1)th week n+1 to obtain ∠α cy , ∠α cy = ∠α n+1 - ∠α n . Output ∠α cy as the bovine back physical sign change data.

[0020] It should be noted that: when ∠α cy is positive, it is the weight gain state, and when ∠α cy is negative, it is the weight loss state; The above steps can accurately determine the growth condition of beef cattle in the past week through the bovine back physical sign change data; The health classification module S200 is used to collect the beef cattle diet data from the nth week to the (n + 1)th week, and perform pathological determination on the beef cattle diet data and the bovine back physical sign change data based on a preset classification model to obtain the health label of the beef cattle. The health label includes "healthy beef cattle", "sub-healthy beef cattle", and "diseased beef cattle"; Specifically, the beef cattle diet data includes the average daily food intake, the average daily water intake, and the average daily rumination frequency; It should be noted that the beef cattle diet data is obtained by averaging the cumulative sum of the daily food intake, daily water intake, and daily rumination frequency over seven days. Among them, the daily food intake and water intake are obtained through pressure sensors installed at the bottom of the food trough and water trough, and the rumination frequency is collected through a frequency sensor worn around the cattle's neck; Specifically, the construction logic of the preset classification model is as follows: Obtain historical beef cattle classification data, and divide the historical beef cattle classification data into a training set and a test set. The historical beef cattle classification data includes beef cattle diet data, changes in the physical signs of the cattle's back, and their corresponding health labels; Specifically, the generation logic of the health label is as follows: Compare the average daily food intake, average daily water intake, average daily rumination frequency of the beef cattle, and the change data of the physical signs of the cattle's back ∠α cy with their corresponding preset normal ranges. If the average daily food intake, average daily water intake, average daily rumination frequency, and the change data of the physical signs of the cattle's back ∠α cy are all within the corresponding preset normal ranges, then mark the corresponding beef cattle as "healthy beef cattle"; If one of the average daily food intake, average daily water intake, average daily rumination frequency, and the change data of the physical signs of the cattle's back ∠α cy is not within the preset normal range, then mark it as "sub-healthy beef cattle"; If two or more of the average daily food intake, average daily water intake, average daily rumination frequency, and the change data of the physical signs of the cattle's back ∠α cy are not within the preset normal range, then mark it as "diseased beef cattle"; It should be noted that the preset normal ranges corresponding to the average daily food intake, average daily water intake, average daily rumination frequency, and the change data of the physical signs of the cattle's back ∠α cy of the beef cattle are determined through historical breeding data; Configure an initial classifier, use the beef cattle diet data and the change data of the physical signs of the cattle's back in the training set as the input data of the initial classifier, and use the corresponding health labels in the training set as the output data of the initial classifier. Train the initial classifier to obtain an initial classification network; Verify the initial classification network through the test set, and output the initial classification network with an accuracy greater than or equal to the preset test accuracy as the pre-constructed classification model.

[0021] The disease prediction module S300 is used to obtain the environmental data of the environment where the beef cattle are located from the nth week to the n + 1th week, and input the diet data and environmental data of individuals with the health label of "sub-healthy beef cattle" into a preset regression model to obtain the disease probability of the beef cattle from the n + 1th week to the n + 2th week; Specifically, the environmental data includes a temperature change trend graph, a humidity change trend graph, and a wind speed change trend graph of the breeding farm; It should be noted that the above data are obtained through a temperature sensing component, a humidity sensing component, and a wind speed sensing component respectively. Among them, the temperature change trend graph, the humidity change trend graph, and the wind speed change trend graph of the breeding farm are automatically generated by Python and Matplotlib; Specifically, the construction logic of the regression model is as follows: Obtain historical regression data, and divide the historical regression data into a test set and a training set. The historical regression data includes environmental data, diet data, and their corresponding disease occurrence probabilities; Specifically, the acquisition logic of the disease occurrence probability is as follows: Obtain historical data for calculating the disease occurrence probability. The historical data for calculating the disease occurrence probability includes the weight change value, the ammonia concentration in the cowshed, and the moisture content of the dry food; Compare the weight change value, the ammonia concentration in the cowshed, and the moisture content of the dry food with their corresponding normal thresholds, and generate a disease occurrence probability based on the corresponding set weight factors. The calculation formula for the disease occurrence probability is: ι = ×γ1 + ×γ2 + ×γ3; Among them, ι is the disease occurrence probability, γ1 + γ2 + γ3 = 1.24, Wc is the weight change value, Nc is the ammonia concentration in the cowshed, and Md is the moisture content of the dry food; Wy is the normal threshold of the weight change value, Ny is the normal threshold of the ammonia concentration in the cowshed, and My is the normal threshold of the moisture content of the dry food; γ1, γ2, and γ3 are the weight factors of the weight change value, ammonia concentration, and moisture content of the dry food respectively, and γ1 + γ2 + γ3 = 1.24; It should be noted that a large change in the weight change value indicates that the beef cattle are rapidly gaining weight or losing weight. Weight gain may lead to the formation of internal tumors, while weight loss indicates that the beef cattle are sick, suffering from anorexia or having problems with the digestive tract; The ammonia concentration is directly affected by the feces of beef cattle. When beef cattle have diarrhea due to diseases (such as rumen acidosis, intestinal infection) and the moisture content of the excrement increases, it will accelerate the decomposition of urea, resulting in a short-term increase in the local ammonia release amount; in addition, some metabolic diseases of beef cattle (such as ketosis) will change the urine pH value and affect the urea decomposition rate; sick cattle will reduce standing activities due to pain or weakness, resulting in the concentration of excrement in the lying area, increasing the local ammonia concentration. Then, the ventilation situation in the cowshed will directly affect the ammonia concentration; Changes in ambient temperature and humidity can significantly affect the humidity stability of forage, thereby affecting the moisture content, palatability, and safety of dry feed. In an environment with high temperature and humidity fluctuations, it is easier for forage to absorb moisture and mildew, leading to a decrease in feed intake, health problems, and even poisoning. γ1, γ2, and γ3 are determined by adjustment after comparison of parameters by the breeding staff based on historical data. The weight change values are regularly recorded by an automatic electronic scale installed in the cattle body weighing channel and uploaded to the system. The ammonia concentration in the cowshed is detected in real time by ammonia sensors deployed in different areas of the cowshed, and data is aggregated in combination with the environmental monitoring terminal. The moisture content of dry feed is measured by sampling with a near-infrared moisture detector before feed feeding, or measured in real time through a moisture detection module built into the automatic feeding system. All data is automatically collected and stored through the Internet of Things platform. Construct a regression network, use the environmental data and diet data in the training set as the input of the regression network, and use the corresponding incidence probability in the training set as the output data of the regression network to obtain an initial regression network. Verify the initial regression network through the test set, and output the initial regression network with a preset test error less than or equal to as the regression model.

[0022] In this embodiment, by collecting the top-down view of the cattle's back for two consecutive weeks to extract the data on the changes in the signs of the cattle's back, and combining the average daily feed intake, daily water intake, and daily rumination frequency of the beef cattle during the same period, a classification model is established to determine the health status of the beef cattle, realizing the accurate distinction of individuals of "healthy beef cattle", "sub-healthy beef cattle", and "diseased beef cattle", which helps to timely identify individuals in abnormal states and manage them, and further realizes the determination of the physiological health status of beef cattle during growth when measuring the body condition of beef. For beef cattle identified as "sub-healthy", further collect the data on the changes in temperature, humidity, and wind speed during the corresponding period, and input them into the regression model in combination with the diet record to calculate the incidence probability for the next week; thus realizing the early prediction of the incidence risk of key individuals, so as to carry out targeted early intervention and treatment in breeding management and avoid the occurrence of a large number of sick beef cattle.

[0023] Real-time Example 2 To implement the execution of a beef cattle body condition information measurement system in Example 1, this embodiment proposes a beef cattle body condition information measurement device, including the following components: An image acquisition component, arranged at the top of the channel for detecting the body condition of beef cattle, for acquiring the top-down view of the beef cattle's back. A diet data acquisition component, including an eating detection sensor arranged at the bottom of the food bowl, a water intake detection sensor arranged at the bottom of the water trough, and a frequency sensor arranged on the cattle's neck, for acquiring the feed intake, water intake, and rumination frequency of individual beef cattle. The environmental data acquisition component includes a temperature sensor, a humidity sensor, and a wind speed sensor installed in the cowshed, which are used to collect the temperature, humidity, and wind speed in the cowshed; The trend chart generation unit is used to read the collected temperature, humidity, and wind speed, and generate corresponding temperature change trend charts, humidity change trend charts, and wind speed change trend charts; The central processing unit calls the classification model and the regression model to perform data analysis on the collected data, and outputs the health label and the disease probability to the display unit; The storage unit is used to store various types of collected data and respond to the trend chart generation unit and the central processing unit; The display unit is used to receive the health label and the disease probability output by the central processing unit in real time and display them on the mobile terminal.

[0024] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired network or a wireless network. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more collections of available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0025] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in the present invention can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professionals can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0026] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0027] In several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only for a beef cattle body condition information measurement system. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0028] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0029] In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0030] As described above, the above are only specific implementation manners of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention and should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

[0031] Finally: The above are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A beef cattle body condition information measurement system, characterized in that, The system includes: A physical condition measurement module: used to compare the top-down views of the cow's back collected in the nth week and the (n + 1)th week to obtain data on the changes in the physical signs of the cow's back; A health classification module, used to collect the beef cattle diet data from the nth week to the (n + 1)th week, and perform pathological determination on the beef cattle diet data and the data on the changes in the physical signs of the cow's back based on a preset classification model to obtain the health label of the beef cattle, and the health label includes "healthy beef cattle", "sub-healthy beef cattle" and "diseased beef cattle"; An onset prediction module, used to obtain the environmental data from the nth week to the (n + 1)th week, and input the diet data and environmental data of the individuals with the health label of "sub-healthy beef cattle" into a preset regression model to obtain the onset probability of the beef cattle from the (n + 1)th week to the (n + 2)th week.

2. The beef cattle body condition information measurement system according to claim 1, wherein The acquisition logic of the data on the changes in the physical signs of the cow's back is as follows: Identify the key points of the midline of the cow's spine and the key points of the cow's tail part in the top-down view of the cow's back, and extract the edge point set M of the outer contour of the body in the top-down view of the cow's back. The edge point set M includes m edge points, and m is an integer greater than 0; Connect the key points of the midline of the cow's spine to obtain the cow's spine curve, respectively draw the nearest perpendicular lines from the m edge points to the cow's spine curve, select the longest nearest perpendicular line, and record the corresponding edge point as P1. Establish a multi-segment plane coordinate system with the cow's spine curve as the y-axis, and obtain the edge point P2 through edge point screening; Connect the key points of the cow's tail part to obtain the cow's tail curve, use the intersection point of the cow's tail curve and the cow's spine curve as the connection point P3, connect the line segments in the order of P1, P3, P2, and do not connect between P1 and P2 to obtain ∠α; Calculate the difference between ∠α in the nth week n and ∠α in the (n + 1)th week n+1 to obtain ∠α cy , where ∠α cy = ∠α n+1 - ∠α n . Output ∠α cy as the data of the change in the cattle back sign.

3. The beef cattle body condition information measurement system according to claim 2, wherein The logic of the edge point screening is as follows: Arrange the m nearest perpendicular lines in descending order of length, select the first two items and mark them as the longest perpendicular line i(1) and the second longest perpendicular line i(u), where u≥2 and u is an integer; Establish a multi-segment plane coordinate system with the cow's spine curve as the y-axis, record the coordinates of the edge point P1 corresponding to the longest perpendicular line i(1) as (X1, Y1), and record the coordinates of the edge point corresponding to the second longest perpendicular line i(u) as (Xu, Yu); Compare X1 with Xu. If X1 < 0 and Xu < 0, or X1 > 0 and Xu > 0, it is determined that the longest perpendicular line i(1) and the second longest perpendicular line i(u) are on the same side, and go to step S124; If X1 < 0 and Xu < 0, or X1 > 0 and Xu > 0, it is determined that the longest i(1) and the second longest perpendicular line i(u) are on different sides, and output the edge point of the second longest perpendicular line i(u) as P2; Select the nearest perpendicular line whose length ranking is second only to the second longest perpendicular line as the alternative perpendicular line, mark it as i(u + 1), let the second longest perpendicular line i(u)= the alternative perpendicular line i(u + 1), and go to step S122.

4. The beef cattle body condition information measurement system according to claim 3, wherein The beef cattle diet data includes the average daily food intake, the average daily water intake, and the average daily rumination frequency.

5. The beef cattle body condition information measurement system according to claim 4, wherein, The construction logic of the preset classification model is as follows: Obtain historical beef cattle classification data, and divide the historical beef cattle classification data into a training set and a test set. The historical beef cattle classification data includes beef cattle diet data, data on the changes in the physical signs of the cow's back, and their corresponding health labels; Configure an initial classifier, use the beef cattle diet data and the data on the changes in the signs of the cattle's back in the training set as the input data of the initial classifier, and use the corresponding health labels in the training set as the output data of the initial classifier. Train the initial classifier to obtain an initial classification network. Verify the initial classification network through a test set, and output the initial classification network with a test accuracy greater than or equal to the preset test accuracy as a pre-built classification model.

6. The beef cattle body condition information measurement system according to claim 5, characterized in that, The generation logic of the health label is as follows: The average daily feed intake, water intake, rumination frequency and back signs of beef cattle were analyzed. cy Compare with the corresponding preset normal range, if the average daily food intake, average daily water intake, average daily rumination frequency and the change data of the cow's back signs ∠α cy If all are within the corresponding preset normal range, the corresponding beef cattle will be marked as "healthy beef cattle"; If the mean daily food intake, mean daily water intake, mean daily rumination frequency, and the data of the change in the cattle back sign ∠α cy are not within the preset normal range for one of them, it is marked as "sub-healthy beef cattle"; If the mean daily food intake, mean daily water intake, mean daily rumination frequency, and the data of the change in the cattle back sign ∠α cy among two or more of them are not within the preset normal range, it is marked as "diseased beef cattle".

7. The beef cattle body condition information measurement system according to claim 6, characterized in that The environmental data includes a temperature change trend graph, a humidity change trend graph, and a wind speed change trend graph of the breeding farm.

8. The beef cattle body condition information measurement system according to claim 7, characterized in that, The construction logic of the regression model is as follows: Obtain historical regression data, and divide the historical regression data into a test set and a training set. The historical regression data includes environmental data, diet data, and their corresponding disease incidence probabilities. Construct a regression network, use the environmental data and diet data in the training set as the input of the regression network, and use the corresponding disease incidence probability in the training set as the output data of the regression network to obtain an initial regression network. Verify the model of the initial regression network through a test set, and output the initial regression network with a test error less than or equal to the preset test error as a regression model.

9. The beef cattle body condition information measurement system according to claim 8, wherein The acquisition logic of the disease incidence probability is as follows: Obtain historical data for calculating the disease incidence probability. The historical data for calculating the disease incidence probability includes the value of weight change, the ammonia concentration in the cowshed, and the moisture content of the dry food. Compare the value of weight change, the ammonia concentration in the cowshed, and the moisture content of the dry food with the corresponding normal thresholds, and generate a disease incidence probability based on the corresponding set weight factors. The calculation formula for the disease incidence probability is: ι = ×γ1 + ×γ2 + ×γ3; where ι is the disease incidence probability, γ1 + γ2 + γ3 = 1.24, Wc is the value of weight change, Nc is the ammonia concentration in the cowshed, and Md is the moisture content of the dry food. Wy is the normal threshold of the value of weight change, Ny is the normal threshold of the ammonia concentration in the cowshed, and My is the normal threshold of the moisture content of the dry food. γ1, γ2, and γ3 are the weight factors of the value of weight change, ammonia concentration, and moisture content of the dry food respectively, and γ1 + γ2 + γ3 = 1.

24.

10. A beef cattle body condition information measurement device for implementing the beef cattle body condition information measurement system according to any one of claims 1-9, characterized in that, It includes the following components: An image acquisition component, which is set at the top of the channel for detecting the physical condition of beef cattle, and is used to acquire a top-down view of the beef cattle's back. A diet data acquisition component, which includes a feeding detection sensor set at the bottom of the food basin, a water intake detection sensor set at the bottom of the water trough, and a frequency sensor set on the cattle's neck, and is used to acquire the feeding amount, water intake amount, and rumination frequency of individual beef cattle. An environmental data acquisition component, which includes a temperature sensor, a humidity sensor, and a wind speed sensor set in the cowshed, and is used to acquire the temperature, humidity, and wind speed in the cowshed. A trend graph generation unit, which is used to read the collected temperature, humidity, and wind speed, and generate corresponding temperature change trend graphs, humidity change trend graphs, and wind speed change trend graphs. A central processing unit, which calls the classification model and the regression model to perform data analysis on the collected various data, and outputs the health label and the disease incidence probability to the display unit. A storage unit, which is used to store various types of collected data and respond to the trend graph generation unit and the central processing unit. A display unit, which is used to receive in real time the health label and the disease incidence probability output by the central processing unit and display them on the mobile terminal.

Images