Cow sign information acquisition method, system and device based on multi-source perception and medium

Abstract

The invention provides a dairy cow sign information acquisition method, system and device based on multi-source perception and a medium, and belongs to the technical field of agricultural informationization, and the method comprises the steps: arranging edge nodes and data acquisition devices in a dairy cow breeding area; the edge node collects the identity label of a dairy cow through the RFID reader-writer, obtains the weight, body shape, body temperature and breast image feature data of the dairy cow through the data collection device, carries out the preprocessing and feature extraction of the collected data, and then uploads the data to the cloud server. The cloud server performs fusion analysis on the processed feature data, extracts the health state and production performance indexes of the dairy cow, and stores the feature data and analysis results in a database; and the cloud server responds to a query request of the user and returns the corresponding feature data and the analysis result to the user. The dairy cow sign data is collected through the multi-source equipment, and is cooperatively processed and analyzed through the edge node and the cloud, so that the breeding management efficiency is improved, and intelligent breeding is realized.

Description

Technical Field

[0001] This application belongs to the field of agricultural informatization technology, and particularly relates to a method, system, device and medium for collecting dairy cow physical sign information based on multi-source perception. Background Art

[0002] In traditional dairy cow breeding, the collection of physical sign information mainly relies on manual observation and simple measurement. The traditional method is not only time-consuming and laborious, but also difficult to guarantee the accuracy and timeliness of data. Breeders need to check each dairy cow one by one and record key physical sign information such as its weight, body temperature, and udder condition. This process is not only inefficient, but also prone to data deviation due to human factors. In addition, due to the lack of real-time monitoring means, many health problems are often discovered only when the symptoms are obvious, missing the best intervention opportunity and seriously affecting the health and milk production of dairy cows.

[0003] With the rapid development of technologies such as the Internet of Things, image recognition, and big data, the application of multi-source perception technology in the agricultural field is becoming increasingly widespread, providing the possibility for the intelligent and automated collection of dairy cow physical sign information. However, most of the current solutions have the following deficiencies: First of all, existing intelligent collection systems mostly focus on a certain type of data, such as only collecting body temperature or weight through sensors, and cannot comprehensively obtain the multi-dimensional physical sign information of dairy cows. The health and production performance of dairy cows are comprehensively affected by multiple factors, and single-dimensional data is difficult to fully reflect their true state. Secondly, even if a certain amount of data can be collected, the data processing ability of existing systems is often insufficient, and it is unable to analyze and feedback the health status of dairy cows in real time. Thirdly, most existing systems stay at the level of data collection and simple statistics, lacking in-depth intelligent analysis functions. It is unable to provide strong support for the health management, breed improvement, production performance evaluation, and disease warning of dairy cows. Summary of the Invention

[0004] In a first aspect, an embodiment of this application provides a method for collecting dairy cow physical sign information based on multi-source perception, including the following steps: S1. Arrange edge nodes and data collection devices in the dairy cow breeding area; the data collection devices include RFID readers, 3D scanning sensors, intelligent electronic scales, infrared thermal imagers, and cameras; S2. The edge nodes collect the dairy cow identity identification through the RFID readers, obtain the dairy cow weight, body shape, body temperature, and udder image feature data through the data collection devices, and then preprocess and extract features from the collected data and upload them to the cloud server; S3. The cloud server performs fusion analysis on the processed feature data, extracts the dairy cow health status and production performance indicators, and saves each feature data and analysis results to the database; S4. The cloud server responds to the user's query request and returns the corresponding feature data and analysis results to the user.

[0005] Further, in step S2, the edge node obtains the dairy cow identity identifier by reading the RFID electronic ear tag of the dairy cow through an RFID reader, obtains the dairy cow weight data through an intelligent electronic scale, obtains the dairy cow 3D point cloud data through a 3D scanning sensor, collects the dairy cow infrared image through an infrared thermal imager, and collects the dairy cow image data through a camera. The limitation of the specific method for the data acquisition device to obtain data provides a standardized data source for subsequent feature extraction and analysis.

[0006] Further, in step S2, the edge node uses a lightweight AI model to preprocess and extract features from the collected data. The specific steps include: Filter the dairy cow weight data and remove outliers; Perform temperature calibration and region segmentation on the dairy cow infrared image to obtain the dairy cow body temperature data; Perform noise reduction on the dairy cow 3D point cloud data and use the ICP algorithm to calculate the body shape parameters to obtain the dairy cow body shape data; Use the lightweight YOLOv5 model as an image recognition model to extract breast features from the dairy cow image data; Perform time synchronization marking on each collected data and associate it with the dairy cow identity identifier. Using a lightweight AI model to preprocess and extract features from the collected data reduces the consumption of computing resources while ensuring the processing effect and improves the data processing efficiency; adopting targeted processing methods for different types of data, such as filtering the weight data and performing temperature calibration and region segmentation on the infrared image, can effectively extract the key features in the data and provide data support for subsequent analysis and evaluation.

[0007] Further, the specific steps for performing time synchronization marking on each collected data and associating it with the dairy cow identity identifier are as follows: Record the timestamp t0 when the RFID reader first reads the dairy cow identity information; Bind the data first collected by each data acquisition device to the dairy cow identity identifier; When the subsequently collected data is within the time window t0±Δt and meets the continuity condition, it is automatically associated with the same dairy cow identity identifier; the continuity condition is that the fluctuation amplitude of each collected data does not exceed a preset threshold. Performing time synchronization marking and identity association on the collected data solves the consistency problem of multi-source data in the time dimension, ensures that the data of the same dairy cow can accurately correspond; by setting the time window and continuity condition, the accuracy and reliability of data association are improved, data confusion is avoided, and accurate data input is provided for subsequent data analysis and model evaluation.

[0008] Furthermore, the specific steps of step S3 are as follows: S31. The cloud server inputs the processed feature data into the health assessment model to perform health status assessment; S32. The cloud server inputs the processed feature data into the production performance evaluation model to perform production performance evaluation; S33. The cloud server saves the health status assessment results, production performance assessment results and various characteristic data to the database. The cloud server evaluates the processed characteristic data through the health assessment model and production performance assessment model, and realizes the quantitative analysis of the health status and production performance of dairy cows; the assessment results and characteristic data are saved in the database to facilitate data management and query, and provide a basis for breeding decisions.

[0009] Furthermore, step S3 also includes the following steps: S34. The cloud server periodically uses the processed feature data to perform incremental training on the health assessment model and the production performance assessment model; S35. The cloud server sends the health assessment model and production performance assessment model that have completed incremental training to the edge node for local health assessment and production performance assessment. Regularly perform incremental training on the health assessment model and production performance assessment model, and send the trained model to the edge node, so that the model can continuously adapt to changes in the cow's vital signs data, improving the accuracy and adaptability of the model; the edge node can perform local evaluation, reducing the frequency of data upload to the cloud, reducing network transmission pressure, and improving the response speed of the system. The cloud server can respond to query requests initiated by users in a variety of ways, and perform corresponding processing based on the query type, providing users with flexible and convenient data query services; users can obtain the characteristic data and analysis results of dairy cows according to actual needs, so as to timely grasp the health and production status of dairy cows and optimize breeding management. Furthermore, the specific steps of step S4 are as follows: S41. The cloud server responds to the query request initiated by the user through the Web terminal, mobile terminal or API; S42. The cloud server parses the query request and determines the query target, data type, and visualization form; S43. The cloud server obtains the cow identity associated with the query target and obtains the corresponding feature data; If it is a real-time query, the health assessment model and production performance assessment model are used to perform real-time health assessment and production performance assessment, and the characteristic data and real-time analysis results are returned to the user; If it is a historical query, historical feature data and historical analysis results are obtained from the database and returned.

[0010] Second aspect, the embodiments of the present application further provide a cow physical sign information acquisition system based on multi-source perception, including: Data acquisition devices, arranged in the cow breeding area, for acquiring cow weight, body shape, body temperature and breast image feature data; Edge nodes, arranged in the cow breeding area, for preprocessing and feature extraction of the acquired data, and uploading to the cloud server; Cloud server, for performing fusion analysis on the processed feature data, extracting cow health status and production performance indicators, saving each feature data and analysis results to the database, and responding to the user's query request, returning the corresponding feature data and analysis results to the user.

[0011] Third aspect, the embodiments of the present application further provide an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps of the cow physical sign information acquisition method based on multi-source perception as described in the first aspect.

[0012] Fourth aspect, the embodiments of the present application further provide a storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the cow physical sign information acquisition method based on multi-source perception as described in the first aspect.

[0013] From the above technical solutions, it can be seen that the present application has the following advantages: In the cow physical sign information acquisition method, system, device and medium based on multi-source perception provided by the present application, through multi-source perception, multi-dimensional physical sign information of cow weight, body shape, body temperature and breast image is comprehensively acquired. Compared with the traditional single data acquisition method, it can more comprehensively and accurately reflect the health status and production performance of cows, providing a basis for reasonable breeding; through the preprocessing of edge nodes and cloud servers, real-time acquisition, processing and analysis of data are realized, and health status and production performance indicators are timely fed back, facilitating breeders to quickly discover problems and take measures, improving breeding efficiency; a lightweight AI model is used for data preprocessing and feature extraction, and health assessment models and production performance assessment models are used for in-depth analysis to realize intelligent and automated physical sign information acquisition and analysis, reducing manual intervention, lowering labor costs, and improving the accuracy and consistency of data processing; by uniformly saving the acquired feature data and analysis results to the database, and responding to the user's query request through the cloud server, providing diversified visual displays, facilitating breeders to manage and analyze data, and providing support for breeding decisions. Description of the Drawings

[0014] To more clearly illustrate the technical solutions of this application, the following will briefly introduce the attached drawings required for the description. Obviously, the attached drawings in the following description are only some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other attached drawings can be obtained based on these attached drawings.

[0015] Figure 1 It is a schematic flowchart of the method for collecting dairy cow physical sign information based on multi-source perception of the present invention.

[0016] Figure 2 It is a schematic diagram of the system for collecting dairy cow physical sign information based on multi-source perception of the present invention. Detailed implementation manners

[0017] In the following, the specific steps of the method for collecting dairy cow physical sign information based on multi-source perception will be described in detail, and various embodiments of the present disclosure will be described more comprehensively. The present disclosure can have various embodiments, and adjustments and changes can be made therein. However, it should be understood that there is no intention to limit the various embodiments of the present disclosure to the specific embodiments disclosed herein, but the present disclosure should be understood to cover all adjustments, equivalents, and / or alternative solutions falling within the spirit and scope of the various embodiments of the present disclosure.

[0018] Exemplarily speaking, in the traditional dairy cow breeding mode, the acquisition of dairy cow physical sign information mainly relies on manual observation and simple manual measurement. This method is not only inefficient and consumes a large amount of manpower, but also it is difficult to guarantee the accuracy and real-time nature of the data. The breeding personnel need to check each dairy cow one by one and manually record key physical sign information such as its weight, body temperature, and udder condition. This process is not only time-consuming and laborious, but also prone to data deviation due to human factors. In addition, due to the lack of effective real-time monitoring means, many health problems are often discovered only when the symptoms are obvious, missing the best intervention opportunity, thus seriously affecting the health condition and milk production of dairy cows.

[0019] In recent years, with the rapid development of emerging technologies such as the Internet of Things, image recognition, and big data, the application of multi-source perception technology in the agricultural field has gradually increased, bringing new opportunities for the intelligent and automated collection of dairy cow physical sign information. However, most of the current solutions have the following limitations: First of all, most existing intelligent acquisition systems focus on the acquisition of a certain type of data. For example, they only measure body temperature or weight through sensors, but cannot comprehensively obtain the multi-dimensional physical signs information of dairy cows. The health and production performance of dairy cows are comprehensively affected by various factors, and single-dimensional data is difficult to fully reflect the true health status of dairy cows. Secondly, even if a certain amount of data can be collected, the data processing ability of existing systems is often insufficient, and it is unable to analyze and feedback the health status of dairy cows in real time. The timeliness and accuracy of data are crucial for timely detection and handling of health problems. Thirdly, most existing systems only stay at the level of data collection and simple statistics, lacking in-depth intelligent analysis functions. They cannot provide strong support for the health management, breed improvement, production performance evaluation, and disease early warning of dairy cows, and it is difficult to meet the needs of modern aquaculture for refined management and intelligent decision-making.

[0020] Therefore, there is an urgent need to develop a system that can comprehensively, real-time, and intelligently collect and analyze the multi-dimensional physical signs information of dairy cows, so as to improve the efficiency and benefits of dairy cow breeding, ensure the health of dairy cows, and optimize production performance.

[0021] In view of the above problems, this embodiment provides a method for collecting dairy cow physical signs information based on multi-source perception, which realizes the multi-source collection of dairy cow physical signs information, covering key information such as dairy cow identity identification, weight, body shape, body temperature, and breast images, provides a data basis for subsequent analysis and processing, avoids the limitations of single data collection, and can more comprehensively and accurately reflect the actual situation of dairy cows.

[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0023] Please refer to Figure 1 The following is a flowchart of a method for collecting dairy cow physical signs information based on multi-source perception in a specific embodiment. The method includes the following steps: S1. Arrange edge nodes and data collection devices in the dairy cow breeding area; the data collection devices include RFID readers, 3D scanning sensors, intelligent electronic scales, infrared thermal imagers, and cameras; It should be noted that by arranging edge nodes and a variety of data collection devices in the dairy cow breeding area, a distributed data collection network is constructed, which can collect the physical signs information of dairy cows in real time and comprehensively; the setting of edge nodes reduces the delay of data transmission and improves the timeliness of data collection; the combined use of a variety of data collection devices ensures the diversity and integrity of the collected data, providing data support for subsequent analysis and processing; S2. The edge node collects the dairy cow identity identification through an RFID reader, obtains the dairy cow weight, body shape, body temperature, and breast image feature data through data collection devices, and then preprocesses and extracts features from the collected data and uploads them to the cloud server; It should be noted that the edge node preprocesses and extracts features from the collected data locally, reducing the amount of data uploaded to the cloud server, reducing the network transmission pressure, and at the same time improving the data processing efficiency; through the preprocessing and feature extraction of the data, noise can be removed, key features can be extracted, the data quality can be improved, and a more accurate data basis can be provided for the fusion analysis of the cloud server; S3. The cloud server performs fusion analysis on the processed feature data, extracts the dairy cow health status and production performance indicators, and saves each feature data and analysis results to the database; It should be noted that the cloud server performs fusion analysis on the processed feature data, can comprehensively consider various physical signs information of the dairy cow, and accurately extract the dairy cow health status and production performance indicators; saving the feature data and analysis results to the database realizes the centralized management and storage of data, facilitates users to query and analyze data, and provides data support for the decision-making of dairy cow breeding; S4. The cloud server responds to the user's query request and returns the corresponding feature data and analysis results to the user; It should be noted that the cloud server can respond to the user's query request in a timely manner, return the corresponding feature data and analysis results according to the user's needs, enabling the user to grasp the health and production status of the dairy cow in real time; improving the convenience and timeliness of the user to obtain information, facilitating the user to adjust the breeding strategy according to the actual situation and optimize the breeding management.

[0024] In the whole process of this embodiment from data collection, preprocessing, feature extraction, fusion analysis to result feedback, the intelligent and automatic collection and analysis of dairy cow physical signs information are realized, providing a basis for dairy cow health management and production performance evaluation.

[0025] Further, as a refinement and extension of the specific implementation manner of the above embodiment, in order to fully illustrate the specific implementation process in this embodiment, another method for collecting dairy cow physical signs information based on multi-source perception is provided. This method includes the following steps: S1. Arrange edge nodes and data collection devices in the dairy cow breeding area; the data collection devices include an RFID reader, a 3D scanning sensor, an intelligent electronic scale, an infrared thermal imager, and a camera; S2. The edge node collects the dairy cow identity identification through an RFID reader, obtains the dairy cow weight, body shape, body temperature, and breast image feature data through data collection devices, and then preprocesses and extracts features from the collected data and uploads them to the cloud server; In step S2, the edge node reads the RFID electronic ear tag of the dairy cow through an RFID reader to obtain the identity identifier of the dairy cow, obtains the weight data of the dairy cow through an intelligent electronic scale, obtains the 3D point cloud data of the dairy cow through a 3D scanning sensor, collects the infrared image of the dairy cow through an infrared thermal imager, and collects the image data of the dairy cow through a camera; It should be noted that the limitation on the specific method of obtaining data by the data acquisition device provides a standardized data source for subsequent feature extraction and analysis; In step S2, the edge node uses a lightweight AI model to preprocess and extract features from the collected data. The specific steps include: Filter and remove outliers from the weight data of the dairy cow; The sliding average is used in the filtering process. The specific formula is as follows:

[0026] where N represents the window size, represents the weight data collected at time i, represents the filtered weight data; Based on the standard deviation method, outliers are removed. Calculate the mean μ and standard deviation σ of the filtered weight data, and regard the data outside the range of μ ± kσ as outliers and remove them. Exemplarily, k = 3; Perform temperature calibration and region segmentation on the infrared image of the dairy cow to obtain the body temperature data of the dairy cow; Specifically, temperature calibration is performed according to the gray value G of the pixel points in the infrared image of the dairy cow:

[0027] where, T represents the temperature value, a and b are calibration coefficients, which are obtained by calibrating a blackbody radiation source with a known temperature; Use the threshold segmentation method to perform region segmentation on the infrared image. Taking the threshold as as an example, the segmented binary image is , and use to represent the temperature value of the pixel point in the original infrared image; If is greater than , then takes 1. If is less than or equal to , then takes 0; After denoising the 3D point cloud data of the dairy cow, use the ICP algorithm to calculate the body shape parameters to obtain the body shape data of the dairy cow; Use Gaussian filtering to denoise the 3D point cloud data of the dairy cow; The body shape parameters are calculated using the ICP algorithm as follows:

[0028] where, represents a data point in the reference point cloud P, represents a data point in the point cloud Q to be registered, and T is the transformation matrix that minimizes the distance between P and TQ; The lightweight YOLOv5 model is used as an image recognition model to extract breast features from dairy cow image data; The training process of the image recognition model is as follows: The deployed camera is used to collect side-view and rear-view images of dairy cows; Specifically, the collected images need to cover different lighting conditions, postures, and breast states; Images with a quantity greater than the set threshold are collected and annotated; During the annotation process, the breast area is annotated using a rectangular box, and the breast states are defined using labels 0 and 1, where 0 represents a healthy breast and 1 represents an abnormal breast, such as swelling or damage; After performing geometric transformations such as rotation or scaling on the collected data, the brightness and saturation are adjusted, and occlusion blocks are randomly added for occlusion simulation; Based on the YOLOv5s model, the number of Backbone convolutional layers is reduced, some 3×3 convolutions are replaced with depthwise separable convolutions, and the number of channels is compressed to obtain the lightweight YOLOv5 model; Exemplarily, the number of Backbone convolutional layers is reduced from 24 to 12, and the number of channels in each layer is reduced by 50%; Define the loss function:

[0029] where, represents the classification loss, represents the loss of target existence, represents the CIoU loss, which is used to optimize the position of the bounding box; and and represent the weights of the three losses respectively; where,

[0030] where, is the distance between the center points of the predicted bounding box b and the ground truth bounding box is the length of the diagonal of the minimum enclosing rectangle, is the aspect ratio consistency coefficient, is the aspect ratio balance weight coefficient; is the intersection over union (IoU) between the predicted bounding box b and the ground truth bounding box ; Initialize the learning rate, step size, and maximum number of training epochs, and an optimizer for optimizing model parameters; Exemplarily, initialize the learning rate to 0.01, the step size to 32, the maximum number of training epochs to 100, and the optimizer to the Stochastic Gradient Descent (SDG) optimizer; First, during the training process, load the pre-trained weight parameters first; Then, freeze some layers of the Backbone and only train the detection head to complete fine-tuning; Exemplarily, freeze the first 10 layers of the Backbone; and introduce L1 regularization during training to prune channels with weights < 0.01; Finally, unfreeze all layers and perform joint optimization; Perform time synchronization marking on each piece of collected data and associate it with the dairy cow identity identifier; It should be noted that using a lightweight AI model to preprocess and extract features from the collected data reduces the consumption of computing resources while ensuring the processing effect, and improves the data processing efficiency; Using targeted processing methods for different types of data, such as filtering weight data, calibrating the temperature and segmenting regions of infrared images, can effectively extract key features from the data and provide data support for subsequent analysis and evaluation; The specific steps for performing time synchronization marking on each piece of collected data and associating it with the dairy cow identity identifier are as follows: Record the timestamp t0 when the RFID reader first reads the dairy cow identity information; Bind the data first collected by each collection device to the dairy cow identity identifier; For subsequent collected data within the time window t0 ± Δt and satisfying the continuity condition, automatically associate it with the same dairy cow identity identifier; The continuity condition is that the fluctuation range of each piece of collected data does not exceed a preset threshold; Specifically, t 0 is the timestamp when the RFID reader first reads the dairy cow identity information, t i is the i timestamp of the data collected by the

[0031] If (time window), then associate this data with the dairy cow identity identifier; x i is the data collected by the i th collection device, is the data collected in the previous collection, then the continuity condition can be expressed as:

[0032] Among them, is a preset threshold value; It should be noted that time synchronization marking and identity association are performed on the collected data, which solves the consistency problem of multi-source data in the time dimension and ensures that the data of the same dairy cow can be accurately corresponding; by setting a time window and continuity conditions, the accuracy and reliability of data association are improved, data confusion is avoided, and accurate data input is provided for subsequent data analysis and model evaluation; S3. The cloud server performs fusion analysis on the processed feature data, extracts dairy cow health status and production performance indicators, and saves each feature data and analysis results to the database; the specific steps of step S3 are as follows: S31. The cloud server inputs the processed feature data into the health assessment model for health status assessment; Specifically, a health assessment model is established using an LSTM neural network and trained using the time series data of historical dairy cow body temperature data, weight data change trends, dairy cow breast characteristics, and activity data (analyzed according to dairy cow 3D point cloud data), and the health status and disease types are output; The cloud server outputs intervention measures according to the dairy cow breeding knowledge graph, such as medication and isolation measures; Specifically, input: time series physical sign data

[0033] Among them, represents body temperature, represents weight, represents breast feature score, represents activity amount; LSTM cell calculation process: Forget gate:

[0034] Input gate:

[0035] Cell state update:

[0036] Output gate:

[0037] Health score output:

[0038] Among them, 、 、 are the activation values of the forget gate, input gate, and output gate respectively; is the candidate status; is the unit status; is the hidden status; , , , is the weight matrix; , , , is the bias term; is the Sigmoid activation function; tanh is the hyperbolic tangent activation function; Exemplarily, when H≥0.7, a disease warning is triggered; S32. The cloud server inputs the processed feature data into the production performance evaluation model for production performance evaluation; Specifically, dairy cow body size data, breast characteristics, weight data, body temperature data, and historical milk production data are obtained to construct a dataset; The XGBoost regression model is used as the production performance evaluation model and trained using the dataset; The prediction formula of the XGBoost regression model can be expressed as:

[0039] where, is the predicted value, is the output of the k th decision tree, K is the number of decision trees; The output of each decision tree can be expressed as:

[0040] is the structure of the decision tree, indicating which leaf node the input feature x falls into, is the weight of the leaf node; The loss function during the training process is as follows:

[0041] The production performance evaluation model is trained with the goal of minimizing the observation error and keeping the decision tree structure as simple as possible; Then, the dairy cow body size data, breast characteristics, weight data, body temperature data, and historical milk production data are input into the trained production performance evaluation model for milk production prediction, reproductive efficiency scoring, and feed conversion rate calculation; The cloud server ranks the milk production within the group based on the milk production prediction results, generates suggestions for the best breeding time according to the reproductive efficiency score combined with the feeding knowledge graph, and generates an optimized feed ratio plan according to the calculated feed conversion rate combined with the feeding knowledge graph; S33. The cloud server saves the health status evaluation results, production performance evaluation results, and each characteristic data to the database; It should be noted that the cloud server evaluates the processed characteristic data through the health evaluation model and the production performance evaluation model, realizing the quantitative analysis of the health status and production performance of dairy cows; saving the evaluation results and characteristic data to the database facilitates data management and query, providing a basis for breeding decisions; S34. The cloud server regularly performs incremental training on the health evaluation model and the production performance evaluation model using the processed characteristic data; Specifically, the incremental training is carried out through the following formula:

[0042] where, are the parameters of the old model, are the parameters of the new model, is the old data set, is the new data set, is the loss function, is the prediction function of the model, is the true value; S35. The cloud server distributes the health evaluation model and the production performance evaluation model that have completed incremental training to the edge nodes for local health evaluation and production performance evaluation; It should be noted that regularly performing incremental training on the health evaluation model and the production performance evaluation model and distributing the trained models to the edge nodes enable the models to continuously adapt to the changes in dairy cow physical sign data, improving the accuracy and adaptability of the models; the edge nodes can perform local evaluation, reducing the frequency of data upload to the cloud, reducing the network transmission pressure, and improving the response speed of the system; S4. The cloud server responds to the user's query request and returns the corresponding characteristic data and analysis results to the user; the specific steps of step S4 are as follows: S41. The cloud server responds to the query request initiated by the user through the Web end, mobile end, or API; S42. The cloud server parses the query request to determine the query target, data type, and visualization form; exemplarily, the query target can be single dairy cow query, group query, or specific time period query; the data type can be raw characteristic data, health evaluation, and production performance analysis; the visualization form can be charts, reports, or warning lists; S43. The cloud server obtains the dairy cow identity identifier associated with the query target and obtains the corresponding characteristic data; For real-time queries, use the health evaluation model and the production performance evaluation model to perform real-time health evaluation and production performance evaluation, and return the characteristic data and real-time analysis results to the user; Specifically, call the health assessment model, input the current feature data, output the health risk score, and match the preset threshold to generate a warning suggestion (such as "abnormal body temperature: it is recommended to isolate and observe"); call the production performance assessment model, input the historical data, and output the milk production prediction curve; generate a feeding optimization plan in combination with metadata such as breed and parity; For historical queries, obtain historical feature data and historical analysis results from the database and return them; Exemplarily, the specific display form can output a health heat map according to user needs, color-code the population body temperature distribution (red = high risk); a trend chart, showing the changes of indicators such as body weight and milk production over time; a warning list, with abnormal individuals to be processed sorted by urgency; It should be noted that the cloud server can respond to query requests initiated by users in various ways, and perform corresponding processing according to the query type, providing users with flexible and convenient data query services; users can obtain the feature data and analysis results of dairy cows according to actual needs, facilitating timely understanding of the health and production status of dairy cows and optimizing breeding management.

[0043] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0044] As Figure 2 shown, the following is an embodiment of a dairy cow physical sign information collection system based on multi-source perception provided by an embodiment of the present disclosure. This system and the above-described embodiments of the dairy cow physical sign information collection method based on multi-source perception belong to the same inventive concept. Details not described in detail in the embodiments of the dairy cow physical sign information collection system based on multi-source perception can refer to the embodiments of the above-described dairy cow physical sign information collection method based on multi-source perception.

[0045] The system includes: Data collection devices, arranged in the dairy cow breeding area, for obtaining feature data of dairy cow body weight, body shape, body temperature, and breast images; Edge nodes, arranged in the dairy cow breeding area, for preprocessing and feature extraction of the collected data, and uploading it to the cloud server; A cloud server, for performing fusion analysis on the processed feature data, extracting dairy cow health status and production performance indicators, saving each feature data and analysis results to the database, and responding to user query requests, and returning the corresponding feature data and analysis results to the user.

[0046] Through the interaction and cooperation of the data acquisition device, edge node, and cloud server, this embodiment realizes the full-process automated processing from data acquisition to analysis and feedback, providing support for the intelligent acquisition and analysis of dairy cow vital sign information.

[0047] The method for collecting dairy cow characteristic information based on multi-source perception provided by the embodiments of this application can be applied to electronic devices. Those skilled in the art can understand that the structure of the electronic devices involved in the embodiments of the present invention does not constitute a limitation on the electronic devices. The electronic devices may include more or fewer components than those shown in the figures, or combine certain components, or have different component arrangements. In the embodiments of the present invention, the electronic devices include, but are not limited to, laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic devices may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown in the figures, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of this application described and / or claimed herein.

[0048] The electronic device may include a processor, an external memory interface, an internal memory, a universal serial bus (USB) interface, a charging management module, a power management module, a battery, a wireless communication module, an audio module, a speaker, a microphone, a sensor module, keys, a camera, a display screen, and a SIM card interface, etc.

[0049] It can be understood that the structure schematically shown in the embodiments of this application does not constitute a specific limitation on the electronic device. In other embodiments of this application, the electronic device may include more or fewer components than those shown in the figures, or combine certain components, or split certain components, or have different component arrangements. The components shown in the figures may be implemented in hardware, software, or a combination of software and hardware.

[0050] The processor may include one or more processing units. For example, the processor may include a central processing unit (CPU), an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Among them, different processing units may be independent devices or integrated in one or more processors.

[0051] Among them, the processor may be the nerve center and command center of the electronic device. The controller may generate operation control signals according to the instruction operation code and timing signal to complete the control of fetching and executing instructions.

[0052] A memory may also be provided in the processor for storing instructions and data. In some embodiments, the memory in the processor is a cache memory. This memory may save the instructions or data that the processor has just used or recycled. If the processor needs to use the instruction or data again, it can be directly called from this memory. This avoids repeated accesses, reduces the waiting time of the processor, and thus improves the system efficiency.

[0053] The above-mentioned electronic device implements the technical solution of the method for collecting dairy cow characteristic information based on multi-source perception in the present application by arranging edge nodes and data collection devices in the dairy cow breeding area; the data collection devices include RFID readers, 3D scanning sensors, intelligent electronic scales, infrared thermal imagers, and cameras; the edge nodes collect the identity identifiers of dairy cows through the RFID readers, obtain the body weight, body shape, body temperature, and breast image characteristic data of dairy cows through the data collection devices, and then preprocess and extract features from the collected data and upload them to the cloud server; the cloud server performs fusion analysis on the processed characteristic data, extracts the health status and production performance indicators of dairy cows, and saves each characteristic data and analysis result to the database; the cloud server responds to the user's query request and returns the corresponding characteristic data and analysis result to the user, achieving the comprehensive collection of multi-dimensional physical signs information of dairy cows through multi-source perception technology, accurately reflecting their health and production performance, realizing real-time data collection, analysis, and feedback through the collaborative work of edge nodes and the cloud server, improving the management efficiency of breeding personnel; using lightweight AI models and evaluation models to intelligently and automatically process data, reducing labor costs and improving accuracy; uniformly storing data and providing visual display for convenient decision-making, overall improving breeding efficiency and benefits, and realizing reasonable breeding.

[0054] In the storage medium provided by the present application, there is a program product capable of implementing the method for collecting dairy cow characteristic information based on multi-source perception.

[0055] The method for collecting dairy cow characteristic information based on multi-source perception includes: arranging edge nodes and data collection devices in the dairy cow breeding area; the data collection devices include RFID readers, 3D scanning sensors, intelligent electronic scales, infrared thermal imagers, and cameras; the edge nodes collect the identity identifiers of dairy cows through the RFID readers, obtain the body weight, body shape, body temperature, and breast image characteristic data of dairy cows through the data collection devices, and then preprocess and extract features from the collected data and upload them to the cloud server; the cloud server performs fusion analysis on the processed characteristic data, extracts the health status and production performance indicators of dairy cows, and saves each characteristic data and analysis result to the database; the cloud server responds to the user's query request and returns the corresponding characteristic data and analysis result to the user.

[0056] In some possible implementation manners, the method for collecting dairy cow characteristic information based on multi-source perception of the present disclosure may be implemented in the form of a program product, which includes program code. When the program product runs on a terminal device, the program code is used to cause the terminal device to execute the steps according to various exemplary embodiments of the present disclosure described in the "Exemplary Method" section above of this specification.

[0057] The storage medium of the present disclosure may employ any combination of one or more readable media. The readable media may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0058] The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Thus, the present invention will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for collecting dairy cow physical sign information based on multi-source perception, characterized in that, It includes the following steps: S1. Arrange edge nodes and data collection devices in the dairy cattle breeding area; the data collection devices include RFID readers, 3D scanning sensors, intelligent electronic scales, infrared thermal imagers, and cameras; S2. The edge nodes collect the dairy cattle identity identifiers through the RFID readers, obtain the dairy cattle weight, body shape, body temperature, and breast image feature data through the data collection devices, and then preprocess and extract features from the collected data and upload them to the cloud server; S3. The cloud server performs fusion analysis on the processed feature data, extracts the dairy cattle health status and production performance indicators, and saves each feature data and analysis results to the database; S4. The cloud server responds to the user's query request and returns the corresponding feature data and analysis results to the user.

2. The method for collecting dairy cow physical sign information based on multi-source perception according to claim 1, characterized in that In step S2, the edge nodes obtain the dairy cattle identity identifiers by reading the RFID electronic ear tags of the dairy cattle through the RFID readers, obtain the dairy cattle weight data through the intelligent electronic scales, obtain the dairy cattle 3D point cloud data through the 3D scanning sensors, collect the infrared images of the dairy cattle through the infrared thermal imagers, and collect the dairy cattle image data through the cameras.

3. The method for collecting dairy cow characteristic information based on multi-source perception according to claim 2, wherein In step S2, the edge nodes use a lightweight AI model to preprocess and extract features from the collected data. The specific steps include: Filter and remove outliers from the dairy cattle weight data; Perform temperature calibration and region segmentation on the infrared images of the dairy cattle to obtain the dairy cattle body temperature data; Perform noise reduction on the dairy cattle 3D point cloud data and then use the ICP algorithm to calculate the body shape parameters to obtain the dairy cattle body shape data; Use the lightweight YOLOv5 model as an image recognition model to extract breast features from the dairy cattle image data; Perform time synchronization marking on each collected data and associate it with the dairy cattle identity identifier.

4. The method for collecting dairy cow characteristic information based on multi-source perception according to claim 3, wherein, The specific steps for performing time synchronization marking on each collected data and associating it with the dairy cattle identity identifier are as follows: Record the timestamp t0 when the RFID reader first reads the dairy cattle identity information; Bind the data first collected by each data collection device to the dairy cattle identity identifier; Subsequent collected data within the time window t0±Δt and meeting the continuity condition are automatically associated with the same dairy cattle identity identifier; the continuity condition is that the fluctuation range of each collected data does not exceed the preset threshold.

5. The method for collecting dairy cow characteristic information based on multi-source perception according to claim 3, wherein The specific steps of step S3 are as follows: S31. The cloud server inputs the processed feature data into the health assessment model for health status assessment; S32. The cloud server inputs the processed feature data into the production performance assessment model for production performance assessment; S33. The cloud server saves the health status assessment results, production performance assessment results, and each feature data to the database.

6. The method for collecting dairy cow characteristic information based on multi-source perception according to claim 5, wherein The following steps are also included in step S3: S34. The cloud server regularly performs incremental training on the health assessment model and production performance assessment model using the processed feature data; S35. The cloud server distributes the health assessment model and production performance assessment model that have completed incremental training to the edge nodes for local health assessment and production performance assessment.

7. The method for collecting dairy cow characteristic information based on multi-source perception according to claim 3, wherein The specific steps of step S4 are as follows: S41. The cloud server responds to the query request initiated by the user through the Web end, mobile end, or API; S42. The cloud server parses the query request to determine the query target, data type, and visualization form; S43. The cloud server obtains the cow identity identifier associated with the query target and obtains the corresponding feature data; For real-time queries, use the health assessment model and production performance assessment model to perform real-time health assessment and production performance assessment, and return the feature data and real-time analysis results to the user; For historical queries, obtain historical feature data and historical analysis results from the database and return them.

8. A cow physical sign information acquisition system based on multi-source perception, characterized in that, Including: Data acquisition devices are arranged in the cow breeding area to obtain feature data such as cow weight, body shape, body temperature, and breast images; Edge nodes are arranged in the cow breeding area to preprocess and extract features from the collected data and upload them to the cloud server; The cloud server is used to perform fusion analysis on the processed feature data, extract cow health status and production performance indicators, save each feature data and analysis results to the database, and respond to the user's query request to return the corresponding feature data and analysis results to the user.

9. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored on the memory and executable on the processor. It is characterized in that when the processor executes the program, it implements the steps of the multi-source perception-based cow physical sign information collection method according to any one of claims 1 to 7.

10. A storage medium, on which a computer program is stored, characterized in that, When the computer program is executed by the processor, it implements the steps of the multi-source perception-based cow physical sign information collection method according to any one of claims 1 to 7.

Images