Method and system for early prediction and intelligent breeding of longevity of dairy cow, and computing device

Abstract

The invention provides a dairy cow longevity early prediction and intelligent breeding method and system and a computing device, the method comprises the following steps: collecting multi-modal physiological data of a target dairy cow in an early fetal stage, the multi-modal physiological data comprising body size data of the target dairy cow; the multi-modal physiological data is preprocessed; calculating a longevity score of the target dairy cow according to the preprocessed multi-modal physiological data; and according to the longevity score, generating a corresponding breeding decision. According to the technical scheme, early-stage accurate prediction of the longevity of the dairy cow can be achieved, the breeding cost is reduced, the prediction accuracy of the longevity of the dairy cow is improved, intelligent breeding of the dairy cow is achieved, and technical support is provided for building a high-yield, efficient and sustainable modern dairy cow breeding system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent livestock and artificial intelligence fusion, in particular to a method and system for early prediction of long life of dairy cows and intelligent breeding and a computing device. BACKGROUND

[0002] Currently, the evaluation of "longevity" in dairy cow breeding still highly depends on historical production performance data (such as parity, milk yield, culling record) or disease record.

[0003] However, such methods have significant limitations. For example, cycle lag: it takes 3-5 parities or even culling to determine the longevity of a dairy cow, which cannot support early breeding decisions; data fragmentation: disease, reproduction, and other records are scattered in different systems, making data integration difficult; body size data is not used intelligently: although body height, chest circumference, and other body size indicators are easy to obtain, they are traditionally used only for linear scoring of body type, and no calculable and predictable quantitative model has been established for longevity.

[0004] In addition, existing body size measurement relies on manual tape or static photography, which has low efficiency, high stress, and high subjective error, and cannot meet the needs of large-scale farms for high-throughput, automated, and non-contact data collection.

[0005] Therefore, a technical solution is needed to accurately predict the longevity of dairy cows early, reduce breeding costs, improve the prediction accuracy of dairy cow longevity, and realize intelligent breeding of dairy cows, thereby providing technical support for building a high-yield, efficient, and sustainable modern dairy farming system. SUMMARY

[0006] The present application aims to provide a method and system for early prediction of longevity of dairy cows and intelligent breeding and a computing device, which can accurately predict the longevity of dairy cows early, reduce breeding costs, improve the prediction accuracy of dairy cow longevity, realize intelligent breeding of dairy cows, and provide technical support for building a high-yield, efficient, and sustainable modern dairy farming system.

[0007] According to an aspect of the present application, a method for early prediction of longevity of dairy cows and intelligent breeding is provided, the method comprising: collecting multi-modal physiological data of a target dairy cow in an early parity stage, the multi-modal physiological data comprising body size data of the target dairy cow; preprocessing the multi-modal physiological data; calculating a longevity score of the target dairy cow according to the preprocessed multi-modal physiological data; generating a corresponding breeding decision according to the longevity score.

[0008] According to some embodiments, multimodal physiological data of target dairy cows in early parity stages are collected, including: The multimodal physiological data was acquired at key access points in the pasture using broadband imaging. The multimodal physiological data included a static body size index set and a dynamic body size index set. The target dairy cow's identity auxiliary information is automatically associated using RFID or image recognition.

[0009] According to some embodiments, the preprocessing of the multimodal physiological data includes: normalizing the body size data according to parity stage.

[0010] According to some embodiments, the longevity score of the target dairy cow is calculated based on the preprocessed multimodal physiological data, including: The preprocessed multimodal physiological data is input into the longevity prediction model to obtain the longevity score.

[0011] According to some embodiments, the longevity prediction model is pre-constructed: Collect a multimodal physiological dataset of dairy cows, including a body size dataset; Based on the preset longevity labeling criteria, a binary classification training dataset containing long-lived dairy cows and non-long-lived dairy cows was constructed. The body size dataset is normalized according to parity stage; The binary classification training dataset is sorted and filtered by the random forest algorithm to retain the subset of key features that are most relevant to the longevity label data. A multilayer perceptron network is trained using the aforementioned key feature subset to obtain the longevity prediction model, which is then used to calculate the longevity score of dairy cows.

[0012] According to some embodiments, preprocessing of the multimodal physiological data further includes: The static and dynamic body size index sets of the target dairy cows are collected, and the mean body size feature is calculated using a data normalization algorithm.

[0013] According to some embodiments, based on the longevity score, corresponding breeding decisions are generated, including: If the longevity score of the target dairy cow is greater than or equal to the decision threshold, the target dairy cow is marked as a longevity candidate breeding cow, and the breeding decision recommendation report recommends that it be given priority for breeding and that its nutritional management be strengthened.

[0014] According to some embodiments, generating corresponding breeding decisions based on the longevity score further includes: If the longevity score of the target dairy cow is less than the decision threshold, the target dairy cow is marked as a low-life-risk cow, triggering a health warning, and the breeding decision recommendation report recommends strengthening health screening, disease monitoring, or initiating early culling assessment.

[0015] According to another aspect of the present invention, a system for early prediction and intelligent breeding of longevity in dairy cows is provided, the system comprising: The data acquisition module is used to collect multimodal physiological data of target dairy cows in the early parity stage, including body size data of the target dairy cows. The data processing module is used to preprocess the multimodal physiological data; The longevity scoring module is used to calculate the longevity score of the target dairy cow based on the preprocessed multimodal physiological data. The breeding decision module is used to generate corresponding breeding decisions based on the longevity score.

[0016] According to another aspect of the present invention, a computing device is provided, comprising: Processor; and A memory that stores a computer program, which, when executed by the processor, implements the method as described in any of the preceding methods.

[0017] According to embodiments of the present invention, static and dynamic body size indicators of dairy cows are acquired through non-contact intelligent sensing devices. A data-driven longevity assessment model is constructed to calculate a longevity score. Based on the longevity score, a structured breeding decision recommendation report is automatically generated. This invention enables early and accurate prediction of dairy cow longevity, reduces breeding costs, improves the accuracy of longevity prediction, achieves early identification and precise breeding of high-longevity dairy cows, realizes intelligent breeding of dairy cows, and provides technical support for building a high-yield, efficient, and sustainable modern dairy farming system.

[0018] It should be understood that the above general description and the following detailed description are merely exemplary and do not limit the invention. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below.

[0020] Figure 1 A flowchart illustrating a method for early prediction and intelligent breeding of dairy cow longevity according to an example embodiment is shown.

[0021] Figure 2 A schematic diagram of a system for early prediction and intelligent breeding of dairy cow longevity, according to an example embodiment, is shown.

[0022] Figure 3 A schematic diagram illustrating the deployment of a broadband imaging system according to an example embodiment is shown.

[0023] Figure 4 A block diagram of a computing device according to an exemplary embodiment is shown. Detailed Implementation

[0024] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided so that the invention will be thorough and complete, and the concept of the exemplary embodiments will be fully conveyed to those skilled in the art. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0025] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a full understanding of embodiments of the invention. However, those skilled in the art will recognize that the technical solutions of the invention can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of the invention.

[0026] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0027] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0028] It should be understood that although the terms first, second, third, etc., may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another. Therefore, the first component discussed below may be referred to as the second component without departing from the teachings of the present invention. As used herein, the term "and / or" includes all combinations of any one and more of the associated listed items.

[0029] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of exemplary embodiments, and the modules or processes in the drawings are not necessarily essential for implementing the present invention, and therefore cannot be used to limit the scope of protection of the present invention.

[0030] Currently, the assessment of the key economic trait of "longevity" in dairy cow breeding still relies heavily on historical production performance data, such as parity, cumulative milk production, reasons for culling and time of culling, as well as records of certain diseases (such as mastitis, hoof diseases, and metabolic diseases). While these indicators reflect the survival ability and health status of dairy cows in the herd to some extent, their application has significant limitations.

[0031] First, the evaluation cycle is severely delayed. Traditional methods typically require waiting for an individual to complete 3 to 5 lactations, or even until it is naturally or forcibly culled, before a reliable judgment can be made on its longevity. This "post-hoc verification" model cannot provide effective decision support for early breeding (such as the dwarf or first-calf stage), greatly limiting the speed and efficiency of genetic improvement.

[0032] Secondly, body size data of dairy cows has not yet been applied intelligently and model-wise to predict longevity. Although body size indicators such as height, chest circumference, rump width, and hoof angles are easily obtained in daily management, and existing research has shown that they are closely related to the health, adaptability, and lifespan of dairy cows, these data are currently mostly used only in traditional linear body shape scoring as part of appearance evaluation, and have not been incorporated into calculable and generalizable quantitative prediction models. This prevents a large amount of potential phenotypic information from being transformed into effective breeding criteria.

[0033] Furthermore, existing methods for measuring body size are outdated, relying mainly on manual measurement with tape measures or static photography followed by manual labeling. This is not only cumbersome and inefficient but also prone to subjective errors due to human factors. More importantly, these contact or semi-contact methods often cause stress in dairy cows, affecting the animals, and are difficult to implement for high-throughput, high-frequency data collection in large-scale farms.

[0034] Therefore, this invention proposes a method for early prediction and intelligent breeding of dairy cow longevity, which can achieve accurate early prediction of dairy cow longevity, reduce breeding costs, improve the prediction accuracy of dairy cow longevity, realize intelligent breeding of dairy cows, and provide technical support for building a high-yield, high-efficiency, and sustainable modern dairy farming system.

[0035] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention.

[0036] Figure 1 A flowchart illustrating a method for early prediction and intelligent breeding of dairy cow longevity according to an example embodiment is shown.

[0037] See Figure 1This invention aims to address the problems of outdated, expensive, subjective, and inefficient existing methods for breeding dairy cows to promote longevity. It provides a method for early prediction and intelligent breeding of dairy cow longevity. By employing multimodal physiological data acquisition and a longevity prediction model, it enables early prediction of longevity assessment during the 2nd-3rd calf stage of dairy cows. It automatically and intelligently acquires body size data through non-contact imaging equipment, outputs standardized scores to achieve quantitative breeding, and guides decisions on retention and culling.

[0038] This invention provides a method for accurately predicting longevity in dairy cows. A cow is defined as long-lived if it meets any of the following conditions: it survives to its fourth calf or later, and its final culling is due to normal aging or economic restructuring, rather than health problems; its annual milk production is greater than or equal to 7000 kg in each of three consecutive lactation periods, and it has no history of major metabolic or structural health issues such as hoof disease, ketosis, or reproductive disorders throughout its lifespan. Non-longevity cows are defined as individuals culled before their third calf due to the aforementioned major health problems, or whose milk production performance remains consistently low and fails to reach the economic threshold. The method flow is described below.

[0039] In S101, multimodal physiological data of target dairy cows in the early parity stage were collected.

[0040] According to some embodiments, the multimodal physiological data is acquired at key access points in the pasture using broadband imaging, and the multimodal physiological data includes a static body size index set and a dynamic body size index set.

[0041] According to some embodiments, a broadband imaging system is deployed in the cow aisle or milking parlor. When a target cow passes through, its multimodal physiological data is automatically triggered and collected synchronously. Three frames of images and a 10-second gait video are automatically captured for each cow crossing the aisle. The multimodal physiological data includes: a static body size index set, including height, length, chest circumference, abdominal circumference, rump length, and rump width; and a dynamic body size index set, including gait span and spinal curvature. The spinal curvature is obtained by extracting the spinal contour from continuously captured gait video sequences and calculating the time-series features of its curvature changes.

[0042] According to some embodiments, the multimodal physiological data also includes identity auxiliary information, which is automatically associated with the target dairy cow's identity auxiliary information via RFID or image recognition. The identity and identity auxiliary information are automatically matched and bound to the target dairy cow's unique cow number, current parity, calving date, and age using a radio frequency identification (RFID) reader integrated into the broadband imaging system or a deep learning-based individual image recognition module.

[0043] According to some embodiments, target detection and image recognition are performed on the multimodal physiological data acquired by the broadband imaging using a lightweight YOLO-Nano target detection model and a MobileNetV3 image recognition model.

[0044] According to some implementations, the YOLO-Nano model is used to determine in real time whether a complete individual cow has entered the imaging area, and only triggers high-resolution data acquisition when a complete individual is detected, thereby significantly reducing the storage and transmission overhead of invalid data. The MobileNetV3 model is used to achieve high-precision individual identification and matching by comparing the unique biometric features of the cow's face or markings when there are no RFID tags or the tags are invalid.

[0045] According to some embodiments, from gait video sequences, a real-time object detection model based on the YOLOv8 architecture is used to locate and crop out the region of interest (ROI) containing the cow's back area frame by frame. Within the ROI, a semantic segmentation network accurately segments the centerline pixel coordinates of the spine. The centerline coordinates of the spine in N consecutive frames (N≥10) are fitted into a cubic spline curve, and M points (M≥20) are sampled at equal intervals along the length of the curve. The local curvature at each sampling point is calculated to form a multidimensional curvature vector. The curvature vector is statistically analyzed in the time dimension, and its mean, variance, peak frequency, and energy entropy are extracted as dynamic body size features that ultimately characterize the curvature of the spine.

[0046] According to some embodiments, based on the multimodal physiological data, longevity tag data is assigned to the target dairy cows using a farm life-cycle management database. The longevity tag is assigned to a target dairy cow when it survives to its fourth calf or later and is ultimately culled due to normal aging or economic restructuring, or when its annual milk production exceeds 7000 kg in each of three consecutive lactation periods and it has no significant metabolic or structural health records throughout its lifespan.

[0047] According to some embodiments, the longevity label construction process also includes data cleaning and enhancement. Specific steps include: for missing or outlier values ​​in the historical data of the pasture, a pasture knowledge graph reasoning method based on graph neural networks (GNN) is used to fill in the missing or outlier values; the pasture knowledge graph uses individual dairy cows as nodes and kinship, herd feeding records, and disease transmission paths as edges. Through a message passing mechanism, the known longevity label information of neighboring nodes is used to infer the most likely label of the target node, thereby effectively expanding the number of high-quality training samples.

[0048] In S103, the multimodal physiological data is preprocessed.

[0049] According to some embodiments, the body size data is normalized according to parity stage. The collected static body size index sets and dynamic body size index sets of the target dairy cows are used to calculate the mean body size feature through a data normalization algorithm to eliminate the inherent scale differences caused by different growth and development stages.

[0050] This method can better adapt to the subtle differences in the growth and development of dairy cows under different pastures, breeds, and even feeding and management conditions, thereby improving the model's generalization ability.

[0051] Body size indicators significantly associated with longevity, such as chest circumference and height, were screened and processed. A data normalization algorithm (used in deep learning to eliminate variability in feature values) was employed to address the differences in body size influence at different growth stages. The feature calculation formula is as follows: , , Where u is the mean of the body size characteristics, and X max and X min These represent the maximum and minimum values ​​of the body size characteristic, respectively, X = {X1, X2, ..., X...} n} represents the sample data of body size features.

[0052] In S105, the longevity score of the target dairy cow is calculated based on the preprocessed multimodal physiological data.

[0053] According to some embodiments, the longevity prediction model is pre-constructed. A multimodal physiological dataset of dairy cows is collected, including a body size dataset. Based on a preset longevity label standard, a binary classification training dataset containing long-lived and non-long-lived dairy cows is constructed. The body size dataset is normalized by parity stage. The binary classification training dataset is ranked and filtered by importance using a random forest algorithm, retaining the key feature subset most relevant to the longevity label data. A multilayer perceptron network is trained using the key feature subset to obtain the longevity prediction model for calculating the longevity score of dairy cows.

[0054] According to some embodiments, the preprocessed multimodal physiological data is input into a longevity prediction model to obtain the longevity score. The processed multimodal physiological data and longevity label data are then input into the longevity prediction model. A random forest algorithm is used to rank and filter all input data by importance, retaining the key feature subset most relevant to the longevity label data. This key feature subset is then input into a multilayer perceptron network to output the longevity score of the target dairy cow.

[0055] According to some embodiments, an integrated deep learning architecture is used to construct a longevity prediction model. The construction process includes two stages: Stage 1: Random forest is used to select key body size features (such as chest circumference, hip width, and spinal curvature); Stage 2: A multilayer perceptron model is used, with normalized body size + parity + age as input, and the output is the probability of longevity, predicting a longevity score of 0-100.

[0056] According to some embodiments, the normalized body size feature vector, along with the current parity and age information, is fed into a pre-trained ensemble deep learning longevity prediction model. The model first uses a random forest algorithm to rank and filter all input features by importance, retaining a subset of key features most correlated with the longevity label. Then, this subset of key features is input into a multilayer perceptron network, outputting a continuous value between 0 and 100 as the longevity score of the target cow.

[0057] In S107, a corresponding breeding decision is generated based on the longevity score.

[0058] According to some embodiments, when the longevity score of the target dairy cow is greater than or equal to a decision threshold, the target dairy cow is marked as a longevity candidate breeding cow, and the breeding decision recommendation report recommends prioritizing its retention for breeding and strengthening its nutritional management. When the longevity score of the target dairy cow is less than the decision threshold, the target dairy cow is marked as a low-life-risk cow, triggering a health warning, and the breeding decision recommendation report recommends strengthening health screening, disease monitoring, or initiating early culling assessment.

[0059] According to some embodiments, a structured breeding decision recommendation report is automatically generated based on the longevity score output in step S105. Specifically, if the score is ≥75, the cow is marked as a "high-longevity candidate breeder," and the report recommends prioritizing its retention for breeding, strengthening nutritional management, and conducting genetic value assessment. If the score is <75, the cow is marked as a "low-longevity risk cow," and a health warning is triggered in the report, recommending strengthened hoof health screening, metabolic disease monitoring, or initiating an early culling economic benefit assessment process.

[0060] According to some embodiments, the decision threshold of the longevity score is dynamically optimized by using the long-term economic benefits of the ranch as a reward signal based on a reinforcement learning framework.

[0061] According to some embodiments, the breeding decision recommendations are not limited to binary judgments, but generate a probability distribution containing confidence intervals. Specifically, the integrated deep learning model is designed as a Bayesian neural network whose output is a probability density function of longevity scores; the breeding decision system performs preliminary classification based on the expected value of this probability distribution and adjusts the conservatism of the decision based on its variance (i.e., uncertainty); when the variance is large, the system automatically suggests a second verification measurement of the cow to reduce the risk of misjudgment.

[0062] According to some embodiments, the breeding decision output unit has an adaptive threshold adjustment function. This function is implemented based on a reinforcement learning framework, using the long-term economic benefits of the ranch (such as total milk production, feed conversion ratio, and veterinary costs) as a reward signal to dynamically optimize the decision threshold for longevity scores. For example, during periods of high feed prices, the system may automatically lower the retention threshold to retain more individuals with potential for longevity, thereby reducing the high cost of breeding replacement cattle; conversely, when the market price of culled cattle is high, the system may raise the threshold to accelerate the turnover of inefficient individuals.

[0063] According to embodiments of the present invention, intelligent closed-loop breeding can be achieved, automating the entire process from "data acquisition → model prediction → decision output". The present invention enables early intervention, with assessment completed as early as the second calf, 2-3 years earlier than traditional methods.

[0064] Figure 2 A schematic diagram of a system for early prediction and intelligent breeding of dairy cow longevity, according to an example embodiment, is shown.

[0065] See Figure 2 The system for predicting the longevity of dairy cows and intelligent breeding includes: data acquisition module 01, data processing module 02, longevity scoring module 03, and breeding decision module 04.

[0066] According to some embodiments, the data acquisition module 01 is used to collect multimodal physiological data of target dairy cows in the early parity stage, the multimodal physiological data including the body size data of the target dairy cows; the data processing module 02 is used to preprocess the multimodal physiological data; the longevity scoring module 03 is used to calculate the longevity score of the target dairy cows based on the preprocessed multimodal physiological data; and the breeding decision module 04 is used to generate corresponding breeding decisions based on the longevity score. Figure 2 The system shown is Figure 1 The methods shown correspond to each other, and the same content will not be repeated.

[0067] According to some embodiments, the data acquisition module 01 is deployed at the exit of the milking parlor or a dedicated passage in the ranch. It consists of a broadband imaging device, an RFID reader, and an edge computing gateway, and is used to capture multimodal physiological data of dairy cows in real time and non-contactly complete identity binding.

[0068] According to some embodiments, the data acquisition module 01 acquires data through a broadband imaging system. The broadband imaging system specifically includes: a visible light high-definition camera for capturing color texture images of cows under natural lighting conditions; a mid-infrared light source array for providing uniform illumination in low-light or nighttime environments, ensuring image quality is not affected by ambient light; and a depth sensor for acquiring three-dimensional point cloud data of the cow's body surface. The visible light camera and depth sensor are coaxially aligned in physical space and achieve millisecond-level synchronous data acquisition through hardware trigger signals, thereby fusing and generating a multi-channel fused image containing RGB color information, near-infrared reflectance information, and depth distance information, providing a robust data foundation for subsequent high-precision body size index extraction. For the deployment of the broadband imaging system, see [link to relevant documentation]. Figure 3 .

[0069] According to some embodiments, the data processing module 02 is also used to receive raw multimodal physiological data from the data acquisition module 01, and to label the raw multimodal physiological data of dairy cows that meet the criteria for longevity with longevity tags based on the pasture's full life cycle management data.

[0070] According to some embodiments, the data processing module 02 runs on a local server in the ranch or in the cloud, and is used to perform data cleaning, adaptive normalization, dynamic body size feature extraction and key feature screening on the labeled multimodal physiological data.

[0071] According to some embodiments, the core of the longevity scoring module 03 is an online-updable integrated deep learning model service, which receives preprocessed feature vectors and outputs standardized longevity scores and related uncertainty measures.

[0072] According to some embodiments, the breeding decision module 04 can optionally be deeply integrated with the ranch's existing management information system or mobile application to transform longevity scores into actionable business instructions, including but not limited to generating a "high-value breeding cattle retention list", a "high-risk cattle health intervention work order", and an "economic benefit-driven culling recommendation report".

[0073] This invention presents a non-contact, low-stress breeding system. Utilizing broadband imaging, it reduces human intervention, improves animal welfare, and boasts high accuracy. In practical applications, validated with 1000 dairy cows, the model's accuracy is ≥90%. Furthermore, this system can be widely deployed at low cost, with a single measurement costing less than 5 yuan, making it suitable for large-scale deployment in ranches with thousands or tens of thousands of cows.

[0074] Figure 4 A block diagram of a computing device according to an exemplary embodiment of the present invention is shown.

[0075] like Figure 4 As shown, the computing device 30 includes a processor 12 and a memory 14. The computing device 30 may also include a bus 22, a network interface card 16, and an I / O interface 18. The processor 12, memory 14, network interface card 16, and I / O interface 18 can communicate with each other via the bus 22.

[0076] Processor 12 may include one or more general-purpose CPUs (Central Processing Units), microprocessors, or application-specific integrated circuits, for executing relevant program instructions. According to some embodiments, computing device 30 may also include a high-performance display adapter (GPU) 20 for accelerating processor 12.

[0077] Memory 14 may include a machine system readable medium in the form of volatile memory, such as random access memory (RAM), read-only memory (ROM), and / or cache memory. Memory 14 is used to store one or more programs containing instructions, as well as data. Processor 12 may read the instructions stored in memory 14 to perform the methods described above according to embodiments of the present invention.

[0078] The computing device 30 can also communicate with one or more networks via the DPU smart network interface card 16. The DPU smart network interface card is used for data processing or external communication, and the central processing unit is used for processing data scheduled by the DPU smart network interface card. The DPU smart network interface card includes a root system-on-a-chip (SoC) and multiple interfaces, through which the SoC performs data communication. The SoC includes a processor and a memory, on which a computer program is stored. When the processor runs the computer program stored in the memory, it implements the method according to an embodiment of the present invention.

[0079] Bus 22 can include address bus, data bus, control bus, etc. Bus 22 provides a path for exchanging information between components.

[0080] It should be noted that, in specific implementations, the computing device 30 may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the device described above may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0081] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method. The computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives, as well as magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), network storage devices, cloud storage devices, or any type of medium or device suitable for storing instructions and / or data.

[0082] This invention also provides a computer program product comprising a computer program operable to cause a computer to perform some or all of the steps of any of the methods described in the above method embodiments.

[0083] Those skilled in the art will clearly understand that the technical solutions of the present invention can be implemented by means of software and / or hardware. In this specification, "unit" and "module" refer to software and / or hardware capable of independently performing or cooperating with other components to perform a specific function, wherein the hardware may be, for example, a field-programmable gate array (FPGA), an integrated circuit, etc.

[0084] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0085] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0086] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0087] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0088] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0089] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention.

[0090] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0091] Exemplary embodiments of the present invention have been specifically shown and described above. It should be understood that the present invention is not limited to the detailed structures, arrangements, or implementations described herein; rather, the present invention is intended to cover various modifications and equivalent arrangements contained within the spirit and scope of the appended provisions.

Claims

1. A method for early prediction and intelligent breeding of longevity in dairy cows, characterized in that, The method includes: Collect multimodal physiological data of target dairy cows in the early parity stage, including body size data of the target dairy cows; The multimodal physiological data are preprocessed; The longevity score of the target dairy cow was calculated based on the preprocessed multimodal physiological data; Based on the longevity score, corresponding breeding decisions are generated.

2. The method according to claim 1, characterized in that, Collect multimodal physiological data from target dairy cows in early parity stages, including: The multimodal physiological data was acquired at key access points in the pasture using broadband imaging. The multimodal physiological data included a static body size index set and a dynamic body size index set. The target dairy cow's identity auxiliary information is automatically associated using RFID or image recognition.

3. The method according to claim 1, characterized in that, Preprocessing of the multimodal physiological data includes: The body size data are normalized according to the parity stage.

4. The method according to claim 1, characterized in that, The longevity score of the target dairy cow is calculated based on the preprocessed multimodal physiological data, including: The preprocessed multimodal physiological data is input into the longevity prediction model to obtain the longevity score.

5. The method according to claim 4, characterized in that, Also includes: The longevity prediction model is pre-built: Collect a multimodal physiological dataset of dairy cows, including a body size dataset; Based on the preset longevity labeling criteria, a binary classification training dataset containing long-lived dairy cows and non-long-lived dairy cows was constructed. The body size dataset is normalized according to parity stage; The binary classification training dataset is sorted and filtered by the random forest algorithm to retain the subset of key features that are most relevant to the longevity label data. A multilayer perceptron network is trained using the aforementioned key feature subset to obtain the longevity prediction model, which is then used to calculate the longevity score of dairy cows.

6. The method according to claim 2, characterized in that, Preprocessing of the multimodal physiological data includes: The static and dynamic body size index sets of the target dairy cows are collected, and the mean body size feature is calculated using a data normalization algorithm.

7. The method according to claim 1, characterized in that, Based on the longevity score, corresponding breeding decisions are generated, including: If the longevity score of the target dairy cow is greater than or equal to the decision threshold, the target dairy cow is marked as a longevity candidate breeding cow, and the breeding decision recommendation report recommends that it be given priority for breeding and that its nutritional management be strengthened.

8. The method according to claim 7, characterized in that, Based on the longevity score, corresponding breeding decisions are generated, including: If the longevity score of the target dairy cow is less than the decision threshold, the target dairy cow is marked as a low-life-risk cow, triggering a health warning, and the breeding decision recommendation report recommends strengthening health screening, disease monitoring, or initiating early culling assessment.

9. A system for early prediction and intelligent breeding of longevity in dairy cows, characterized in that, The system includes: The data acquisition module is used to collect multimodal physiological data of target dairy cows in the early parity stage, including body size data of the target dairy cows. The data processing module is used to preprocess the multimodal physiological data; The longevity scoring module is used to calculate the longevity score of the target dairy cow based on the preprocessed multimodal physiological data. The breeding decision module is used to generate corresponding breeding decisions based on the longevity score.

10. A computing device, characterized in that, include: processor; as well as A memory storing a computer program that, when executed by the processor, implements the method as described in any one of claims 1-8.

Images