Neural network-based evaluation method for follicular development degree of sow in lactation period

Abstract

The invention relates to the technical field of livestock breeding, in particular to a sow lactation follicular development degree evaluation method based on a neural network, which comprises the following steps: based on a sow in a lactation period, acquiring physiological parameters of the sow in the lactation period in real time by using various sensors, and measuring and recording body temperature, heart rate and respiratory rate data; and generating a physiological parameter data set. According to the method, the physiological parameters of the sows are monitored and analyzed in real time and associated with follicle maturity analysis, reproduction management is optimized, reproduction opportunity errors caused by data delay and errors are reduced, and the pregnancy rate and production efficiency of the sows are improved by evaluating the maturity and size of follicles and optimizing opportunity selection of artificial insemination; by quantitatively evaluating the influence of nutrient intake on follicular development, a scientific basis is provided for feed management, nutrient supply is optimized, breeding efficiency and sow health are improved, and economic benefits and sustainable development level of animal husbandry are enhanced.

Description

Technical Field

[0001] The invention relates to the technical field of animal husbandry, and in particular to a method for evaluating the degree of follicle development of sows during lactation based on a neural network. Background Art

[0002] The field of animal husbandry technology focuses on improving livestock production efficiency, product quality and animal welfare, including genetic selection, feed optimization, reproduction management, disease prevention and control technologies. It aims to improve animal breeding conditions and management through scientific methods and technologies, combine the development of information technology and biotechnology, and use data analysis, biomarkers and real-time monitoring systems to improve reproductive efficiency and animal health, improve operational efficiency and production sustainability, reduce environmental impact and improve energy efficiency, meet consumer and regulatory requirements, and optimize the breeding environment in combination with research on animal behavior to enhance the overall production performance and quality of life of animals.

[0003] Among them, the evaluation method of the degree of follicular development during sow lactation aims to monitor the development of sow follicles through a variety of methods, optimize reproductive management and improve reproductive efficiency, and evaluate the size, number and developmental status of follicles by utilizing imaging technology and data analysis methods. It helps livestock farmers to accurately grasp the physiological state of sows, reasonably arrange breeding plans, adjust feeding measures and breeding strategies, improve the ovulation rate and overall reproductive efficiency of sows, enhance the production performance of sows and the growth and health of offspring, and optimize economic benefits and sustainable development by conducting artificial insemination and reproductive intervention at the appropriate time.

[0004] The traditional method for evaluating the degree of follicular development during sow lactation relies on regular physical examinations and limited data collection, which is obviously insufficient in terms of real-time and accuracy. Physical examinations cannot provide continuous monitoring, and the non-real-time processing of data leads to delays in the evaluation of the sow's physiological state, affecting the timeliness of reproductive decisions. It is difficult to directly link the nutritional status of sows with follicular development, resulting in inflexible adjustments to feeding strategies that cannot be optimized according to actual physiological needs, limiting the improvement of reproductive efficiency and the reduction of production costs, affecting animal welfare and the sustainable development of the breeding industry, and lacking real-time monitoring and data analysis technology, which affects the adjustment of reproductive strategies and feeding management. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a method for evaluating the degree of follicle development of sows during lactation based on a neural network.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme: a method for evaluating the degree of follicle development of sows during lactation based on a neural network, comprising the following steps:

[0007] S1: Based on lactating sows, multiple sensors are used to collect the physiological parameters of sows during lactation in real time, measure and record body temperature, heart rate and respiratory rate data, and generate physiological parameter data sets;

[0008] S2: Based on the physiological parameter data set, the lactation cycle length data of multiple sows are recorded, and the change data of the sow weight and back fat thickness during the lactation cycle are monitored to generate sow lactation cycle data;

[0009] S3: Based on the sow lactation cycle data, an ultrasonic image of a lactating sow is collected, and the image area of ​​the follicle is segmented to identify the boundary and structural characteristics of the follicle, thereby generating follicle image morphology data;

[0010] S4: extracting and analyzing multi-scale texture features of follicles based on the follicle image morphological data, evaluating the maturity and size of the follicles, and forming follicle measurement data;

[0011] S5: Based on the follicle measurement data, combined with the feed consumption data, the effect of nutrient intake on follicle development is analyzed, and the effect of various physiological parameters of sows on follicle development is quantitatively evaluated, and the correlation analysis results are output;

[0012] S6: Based on the correlation analysis results, by identifying the time dependence in the follicle development pattern, a dynamic prediction model of follicle development is constructed to predict the follicle development status of multiple sows and generate follicle development prediction data.

[0013] As a further scheme of the present invention, the physiological parameter data set includes sow physiological parameter collection records, collection time information, and data preprocessing results; the sow lactation cycle data includes lactation cycle length, weight change data within the cycle, and back fat thickness change data within the cycle; the follicle image morphology data includes follicle image collection results, follicle boundary identification data, and follicle structure feature information; the follicle measurement data includes multi-scale texture features of follicle images, follicle maturity scores, and follicle size measurement data; the correlation analysis results include the influence relationship between nutritional intake and follicular development, the quantitative evaluation results of the relationship between physiological parameters and follicular development, and the quantitative data of the influence of multiple variables on follicular development; the follicle development prediction data includes a dynamic prediction model, prediction results of the follicular development status of multiple sows, and prediction model parameter optimization records.

[0014] As a further solution of the present invention, based on lactating sows, multiple sensors are used to collect the physiological parameters of the sows during lactation in real time, and the body temperature, heart rate and respiratory rate data are measured and recorded. The steps of generating the physiological parameter data set are specifically as follows:

[0015] S101: Based on lactating sows, ultrasonic sensors, infrared sensors, heart rate, respiratory rate and body temperature sensors are used to collect and record the back fat thickness, body temperature, heart rate and respiratory rate data of the sows in real time to generate physiological data collection results;

[0016] S102: Based on the physiological data collection result, formatting the collected physiological parameter data, including unit conversion and data structure standardization, to generate a standardized processing result;

[0017] S103: Based on the standardization processing result, the backfat thickness, body temperature, heart rate, and respiratory rate information are sorted according to the time information of data collection, outliers are identified and removed, data integrity is optimized, and a physiological parameter data set is generated.

[0018] As a further solution of the present invention, based on the physiological parameter data set, the lactation cycle length data of multiple sows are recorded, and the change data of the sow weight and back fat thickness during the lactation cycle are monitored. The steps of generating the sow lactation cycle data are specifically as follows:

[0019] S201: Based on the physiological parameter data set, by collecting the start time point and end time point information of the lactation cycles of multiple sows, recording the lengths of the lactation cycles of multiple sows, and generating lactation cycle record data;

[0020] S202: Based on the lactation cycle recorded data, by monitoring the changes of the sow's weight and backfat thickness in multiple stages during the lactation cycle, the weight and backfat thickness data are grouped according to the stages to generate weight and thickness change data;

[0021] S203: Based on the weight and thickness change data, identify the change pattern of back fat thickness during the lactation cycle, analyze the correlation between the cycle stage and the back fat change, and generate sow lactation cycle data.

[0022] As a further embodiment of the present invention, the specific formula for calculating the correlation between the cycle phase and the backfat change is:

[0023]

[0024] Among them, x i represents the lactation cycle phase of a single observation, represents the average value of the lactation cycle stage, y i represents the corresponding observed change in backfat thickness, represents the mean value of the change in backfat thickness, and r is the correlation coefficient between the two, which is used to quantify the strength and direction of the linear relationship between the lactation cycle stage and the change in backfat thickness.

[0025] As a further solution of the present invention, based on the sow lactation cycle data, ultrasound images of lactating sows are collected, and the steps of segmenting the image area of ​​follicles, identifying the boundaries and structural features of follicles, and generating follicle image morphological data are specifically as follows:

[0026] S301: Based on the sow lactation cycle data, use an ultrasonic device to scan the lactating sow, collect ultrasonic images of follicles, and generate follicle image collection data;

[0027] S302: performing image preprocessing based on the follicle image acquisition data, including adjusting brightness and contrast, optimizing image quality in combination with noise filtering, and generating a preprocessed follicle image set;

[0028] S303: Based on the pre-processed follicle image set, identify and segment the follicle region in the image, analyze and mark the boundary and structural features of the follicle, and generate follicle image morphological data.

[0029] As a further solution of the present invention, based on the follicle image morphological data, the multi-scale texture features of the follicle are extracted and analyzed, the maturity and size of the follicle are evaluated, and the steps of forming the follicle measurement data are specifically as follows:

[0030] S401: extracting texture features of the follicles in the image based on the follicle image morphology data, including texture roughness, directionality and uniformity, and generating follicle texture feature data;

[0031] S402: Based on the follicle texture feature data, analyzing the correlation between texture parameters and follicle maturity, evaluating the correlation between texture features and follicle size, and generating follicle maturity analysis data;

[0032] S403: Based on the follicle maturity analysis data, the maturity and size of follicles of multiple lactating sows are evaluated and recorded to generate follicle measurement data.

[0033] As a further embodiment of the present invention, the specific formula for evaluating the correlation between texture features and follicle size is:

[0034]

[0035] Among them, y i represents the maturity score of the i-th follicle, f i represents the maturity of the i-th follicle predicted based on texture features, represents the average score of all follicles’ maturity, n represents the total number of follicle samples, and R 2 Represents the model explanation, which is used to measure the contribution of texture features to the prediction of follicle maturity.

[0036] As a further solution of the present invention, based on the follicle measurement data, combined with the feed consumption data, the effect of nutrient intake on follicle development is analyzed, and the effect of various physiological parameters of sows on follicle development is quantitatively evaluated. The steps of outputting the correlation analysis results are specifically as follows:

[0037] S501: Based on the follicle measurement data, collecting feed consumption records of multiple sows, evaluating the nutritional intake of multiple sows, and generating feed consumption data;

[0038] S502: Based on the feed consumption data, calculating the correlation between the nutrient intake and the follicle development parameters, including the speed of follicle development and the size of follicles, and generating a nutritional impact assessment result;

[0039] S503: Based on the nutritional impact assessment results, combined with the physiological parameters of multiple sows, including back fat thickness, body temperature, heart rate, and respiratory rate, identify and quantify the relationship between multiple physiological parameters and follicle development, and generate correlation analysis results.

[0040] As a further solution of the present invention, based on the correlation analysis results, by identifying the time dependence in the follicular development pattern, a dynamic prediction model of follicular development is constructed to predict the follicular development status of multiple sows, and the steps of generating follicular development prediction data are specifically as follows:

[0041] S601: Based on the correlation analysis result, by performing time series analysis on the follicle development data, analyzing the periodic data changes of the follicle development state, and generating a time-dependent data set;

[0042] S602: Based on the time-dependent data set and the correlation between multiple physiological parameters of lactating sows and follicle development, a dynamic prediction model of follicle development is constructed, model parameters are adjusted to match actual observation data, and a prediction model adjustment record is generated;

[0043] S603: Based on the prediction model adjustment record, the model is verified by simulating multiple development scenarios, the follicle development status of multiple lactating sows is predicted, and follicle development prediction data is generated.

[0044] Compared with the prior art, the advantages and positive effects of the present invention are:

[0045] In the present invention, by real-time monitoring and analysis of the physiological parameters of sows and associating them with the follicle maturity analysis, the reproduction management is optimized, and the errors in the breeding timing caused by data delays and errors are reduced. By evaluating the maturity and size of follicles, the timing of artificial insemination is optimized, and the conception rate and production efficiency of sows are improved. By quantitatively evaluating the effect of nutrient intake on follicle development, a scientific basis is provided for feed management, the nutrient supply is optimized, the reproductive efficiency and sow health are improved, and the economic benefits and sustainable development level of animal husbandry are enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a schematic diagram of the main steps of the present invention;

[0047] Figure 2 It is a schematic diagram of the refinement of S1 of the present invention;

[0048] Figure 3 It is a schematic diagram of the refinement of S2 of the present invention;

[0049] Figure 4 It is a schematic diagram of the refinement of S3 of the present invention;

[0050] Figure 5 It is a schematic diagram of the refinement of S4 of the present invention;

[0051] Figure 6 It is a detailed schematic diagram of S5 of the present invention;

[0052] Figure 7 It is a detailed schematic diagram of S6 of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0054] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0055] See also Figure 1The present invention provides a technical solution: a method for evaluating the degree of follicle development of sows during lactation based on a neural network, comprising the following steps:

[0056] S1: Based on lactating sows, multiple sensors are used to collect the physiological parameters of sows during lactation in real time, measure and record body temperature, heart rate and respiratory rate data, and generate physiological parameter data sets;

[0057] S2: Based on the physiological parameter data set, the lactation cycle length data of multiple sows are recorded, and the change data of sow weight and back fat thickness during the lactation cycle are monitored to generate sow lactation cycle data;

[0058] S3: Based on the sow lactation cycle data, ultrasound images of lactating sows are collected, and the image area of ​​the follicles is segmented to identify the follicle boundaries and structural features, thereby generating follicle image morphology data;

[0059] S4: Based on the follicle image morphological data, extract and analyze the multi-scale texture features of the follicle, evaluate the maturity and size of the follicle, and form the follicle measurement data;

[0060] S5: Based on the follicle measurement data, combined with the feed consumption data, the effect of nutrient intake on follicle development is analyzed, and the influence of various physiological parameters of sows on follicle development is quantitatively evaluated, and the correlation analysis results are output;

[0061] S6: Based on the results of correlation analysis, by identifying the time dependence in the follicular development pattern, a dynamic prediction model of follicular development was constructed to predict the follicular development status of multiple sows and generate follicular development prediction data.

[0062] The physiological parameter data set includes sow physiological parameter collection records, collection time information, and data preprocessing results. The sow lactation cycle data includes the length of the lactation cycle, weight change data within the cycle, and back fat thickness change data within the cycle. The follicle image morphology data includes follicle image collection results, follicle boundary recognition data, and follicle structure feature information. The follicle measurement data includes multi-scale texture features of follicle images, follicle maturity scores, and follicle size measurement data. The correlation analysis results include the influence of nutritional intake on follicular development, the quantitative evaluation results of the relationship between physiological parameters and follicular development, and the quantitative data of the influence of multiple variables on follicular development. The follicle development prediction data includes dynamic prediction models, prediction results of follicular development status of multiple sows, and prediction model parameter optimization records.

[0063] See also Figure 2 Based on lactating sows, a variety of sensors are used to collect the physiological parameters of sows during lactation in real time, measure and record body temperature, heart rate and respiratory rate data. The specific steps for generating physiological parameter data sets are as follows:

[0064] S101: Based on lactating sows, ultrasonic sensors, infrared sensors, heart rate, respiratory rate and body temperature sensors are used to collect and record the back fat thickness, body temperature, heart rate and respiratory rate data of the sows in real time to generate physiological data collection results;

[0065] In sub-step S101, based on lactating sows, ultrasonic sensors are used to measure back fat thickness, infrared sensors and body temperature sensors jointly record body temperature, heart rate sensors and respiratory rate sensors capture heart rate and respiratory rate data respectively. By integrating multi-source sensing technology, key physiological indicators are monitored and recorded in real time. The target data reflects the health status of the sows. The data time between sensors is synchronized to ensure data consistency. The raw data collected by the sensor undergoes preliminary signal processing, including signal amplification, filtering and digitization, and is converted into an electronic data format for subsequent processing. During the data processing process, the real-time and stability requirements of the data transmission and storage of the real-time data acquisition system are considered. Each data point has a timestamp to facilitate subsequent data tracing and processing, laying the foundation for subsequent data formatting and data set generation.

[0066] S102: Based on the physiological data collection result, formatting the collected physiological parameter data, including unit conversion and data structure standardization, to generate a standardized processing result;

[0067] In sub-step S102, based on the physiological data collection results, data formatting technology is used to convert units and standardize data structures for various physiological parameter data, including using a unified script to convert body temperature from Fahrenheit to Celsius, adjusting heart rate and respiratory rate data according to a preset format, and automated processing through programming languages ​​such as Python to ensure data consistency and availability. The data processing libraries used in the process include Pandas and Numpy for data cleaning and conversion. The standardized data structure facilitates storage and query, reduces the complexity of subsequent processing, and provides a guarantee for establishing an efficient data processing process and data quality control. The output result is a complete set of standardized data files, which prepares a sufficient data foundation for subsequent data sorting and outlier processing.

[0068] S103: Based on the standardized processing results, the backfat thickness, body temperature, heart rate, and respiratory rate information are sorted according to the time information of data collection, outliers are identified and removed, data integrity is optimized, and a physiological parameter data set is generated;

[0069] In sub-step S103, based on the standardized processing results, the physiological parameter data set is further processed using data sorting and outlier processing algorithms. The data is sorted according to the time information of the collection to ensure the temporal correctness of the data sequence. The outlier recognition technology Z-score and IQR method are used to detect and eliminate outliers in the data. The target algorithm is used to identify values ​​that deviate too much or too little from the average level to improve the accuracy and reliability of the data set. The data sorting and outlier processing processes are implemented through specialized data processing software. The processed data set has a high degree of neatness and optimized information integrity, which enhances the application value of the data, lays a solid foundation for data analysis and application, and outputs the physiological parameter data set.

[0070] See also Figure 3 Based on the physiological parameter data set, the lactation cycle length data of multiple sows are recorded, and the change data of sow weight and back fat thickness during the lactation cycle are monitored. The specific steps for generating sow lactation cycle data are as follows:

[0071] S201: Based on the physiological parameter data set, by collecting the start time point and end time point information of the lactation cycles of multiple sows, recording the length of the lactation cycles of the multiple sows, and generating lactation cycle record data;

[0072] In sub-step S201, the start and end time of the lactation cycle of each sow are extracted from the physiological parameter data set. The system uses data labeling technology to record the length of the lactation cycle of each sow, and uses time series analysis technology to determine the exact start and end time points of the cycle. When entering data, a standardized time format is used to ensure the consistency and accuracy of the time data. For the calculation of the length of the lactation cycle, a date difference algorithm is used and implemented in Python's datetime library. The cycle length is obtained by calculating the difference between two dates. The data is crucial for understanding the physiological change cycle of sows. The recorded data after data integration provides a basis for subsequent analysis, ensuring the application value of the data and further in-depth research.

[0073] S202: Based on the lactation cycle record data, by monitoring the changes of sow weight and backfat thickness in multiple stages during the lactation cycle, the weight and backfat thickness data are grouped according to the stages to generate weight and thickness change data;

[0074] In sub-step S202, based on the lactation cycle record data, the changes in sow weight and backfat thickness in different lactation stages are monitored, and the weight and backfat thickness are classified according to the time stage using grouping statistical technology. The grouping process uses the data clustering algorithm K-means to cluster the data points according to the set time stage to form clear groups for trend analysis. The Python programming language is used in conjunction with the Pandas library for data processing to ensure that each group accurately reflects the average weight and backfat thickness of the corresponding stage. The data provides an intuitive perspective for understanding physiological changes and provides data support for studying the physical changes of sows throughout the lactation cycle. The generated weight and thickness change data lays a data foundation for pattern recognition and trend analysis.

[0075] S203: Based on the weight and thickness change data, identifying the change pattern of back fat thickness during the lactation cycle, analyzing the correlation between the cycle stage and the back fat change, and generating sow lactation cycle data;

[0076] The specific formula for calculating the correlation between cycle phase and backfat changes is:

[0077]

[0078] Among them, x i represents the lactation cycle phase of a single observation, represents the average value of the lactation cycle stage, y i represents the corresponding observed change in backfat thickness, represents the mean value of the change in backfat thickness, and r is the correlation coefficient between the two, which is used to quantify the strength and direction of the linear relationship between the lactation cycle stage and the change in backfat thickness.

[0079] formula:

[0080]

[0081] Detailed explanation of the formula and the process of formula calculation and derivation:

[0082] The formula was used to calculate the Pearson correlation coefficient between lactation cycle stage and backfat thickness changes, and the results were used to assess the strength and direction of the relationship between the two;

[0083] Parameter meaning and setting value:

[0084] x i is the specific day number of the lactation cycle stage, which is assumed to be a continuous number from 1 to 5, reflecting the time course of the measurement period. is the average value of the lactation cycle stage, reflects the midpoint of the overall measurement period;

[0085] y iis the back fat thickness measurement value of the corresponding day, assuming the data is [2.1, 2.3, 2.5, 2.5, 2.7] mm, reflecting the specific back fat thickness, is the average backfat thickness, mm, representing the average level of backfat thickness during the measurement period;

[0086] Substitute the parameters into the formula for calculation:

[0087]

[0088] The result r≈0.971 indicated that there was a positive correlation between the lactation cycle stage and the change in backfat thickness, indicating that backfat thickness increased as the lactation cycle progressed. The results were used to adjust feeding strategies and optimize breeding plans.

[0089] See also Figure 4 Based on the sow lactation cycle data, ultrasound images of lactating sows are collected, and the steps of segmenting the image area of ​​follicles, identifying the boundaries and structural features of follicles, and generating follicle image morphological data are as follows:

[0090] S301: Based on the sow lactation cycle data, the sow in lactation period is scanned by using an ultrasonic device to collect ultrasonic images of follicles and generate follicle image collection data;

[0091] In sub-step S301, an ultrasonic device is used to scan a lactating sow and collect ultrasonic images of follicles. The process involves using an ultrasonic sensor to provide clear in vivo imaging and capture the size and shape of the follicles. Image acquisition uses dynamic imaging technology to ensure the stable acquisition of continuous images. The software interface of the ultrasonic device is used to adjust the appropriate scanning depth and gain settings to optimize the clarity and detail of the image. After each scan, the image data is automatically saved in the connected computer system in the format of a DICOM file. The target file records the scanning parameters and image metadata, providing raw data for subsequent image processing. The generated follicle image acquisition data provides the necessary basis for subsequent image preprocessing and analysis.

[0092] S302: performing image preprocessing based on the follicle image acquisition data, including adjusting brightness and contrast, optimizing image quality in combination with noise filtering, and generating a preprocessed follicle image set;

[0093] In sub-step S302, necessary preprocessing operations are performed on the follicle image acquisition data to improve the image quality. The preprocessing includes brightness adjustment and contrast optimization. The target adjustment is performed through the image processing software Adobe Photoshop to ensure the optimal visual effect of the image. A Gaussian filter is used to reduce random noise in the image and maintain the recognition of the follicle structure. Through target technical adjustment, the details of the image are enhanced and the edges are clearer, making subsequent image analysis more accurate and effective. The preprocessed image set is organized into a sequence, and each image is marked, including parameter changes before and after processing. Recording target data is very useful for comparing the effects of different processing techniques. The generated preprocessed follicle image set provides high-quality image resources for identifying and analyzing the morphology of follicles.

[0094] S303: Based on the pre-processed follicle image set, identifying and segmenting the follicle region in the image, analyzing and marking the boundary and structural features of the follicle, and generating follicle image morphological data;

[0095] In the above content, based on the preprocessed follicle image set, the follicle area in the image is identified and segmented. According to the formula Calculate the morphological data of follicles;

[0096] Where S represents the total area of ​​the follicle, p i Represents the indicator function of whether the i-th pixel is a follicle area, a i represents the area of ​​the i-th pixel;

[0097] Detailed explanation of the formula and the process of formula calculation and derivation:

[0098] Assume that the follicle area consists of 100 pixels, the area of ​​each pixel is 0.1 mm2, and the area of ​​each follicle pixel is a i =0.1mm2, all pixels in the follicle area belong to the follicle, p i =1.

[0099] Calculate S:

[0100]

[0101] The result 10mm2 shows that under the given image segmentation parameters and pixel area, the total area of ​​the follicle is 10 square millimeters. The formula is used to accurately measure the total area of ​​the follicle region in the image, providing quantitative data on the size and development status of the follicle. The accuracy and reliability of the results are ensured through precise boundary recognition and area calculation.

[0102] See also Figure 5 Based on the follicle image morphological data, the multi-scale texture features of the follicle are extracted and analyzed to evaluate the maturity and size of the follicle. The specific steps for forming the follicle measurement data are as follows:

[0103] S401: extracting texture features of the follicles in the image based on the follicle image morphology data, including texture roughness, directionality and uniformity, and generating follicle texture feature data;

[0104] In sub-step S401, key texture features are extracted from the follicle image morphological data. Image analysis techniques, including grayscale co-occurrence matrix, are used to calculate texture roughness, directionality, and uniformity. The target features are obtained by analyzing the spatial relationship of the grayscale pixel values ​​of the image. The GLCM method describes the statistical variation of texture from multiple angles to ensure comprehensive texture feature capture. The data is calculated using image processing software MATLAB or OpenCV. The processed data shows subtle structural differences on the surface of the follicle. The target texture data is further encoded and stored to generate follicle texture feature data with quantitative description, providing accurate basic data for subsequent maturity analysis.

[0105] S402: Based on the follicle texture feature data, analyzing the correlation between texture parameters and follicle maturity, evaluating the correlation between texture features and follicle size, and generating follicle maturity analysis data;

[0106] The specific formula for evaluating the correlation between texture features and follicle size is:

[0107]

[0108] Among them, y i represents the maturity score of the i-th follicle, f i represents the maturity of the i-th follicle predicted based on texture features, represents the average score of all follicles’ maturity, n represents the total number of follicle samples, and R 2 Represents the model explanation, which is used to measure the contribution of texture features to the prediction of follicle maturity.

[0109] formula:

[0110]

[0111] Detailed explanation of the formula and the process of formula calculation and derivation:

[0112] The formula was used to calculate the correlation between texture features and follicle maturity;

[0113] Parameter meaning and setting value:

[0114] n represents the total number of follicle samples, which is assumed to be 5;

[0115] y i represents the actual maturity score of the i-th follicle, assuming it is 85, 90, 88, 95, 92, Represents the average score of maturity of all follicles.

[0116] f i represents the maturity of the i-th follicle predicted based on texture features, assumed to be 86, 89, 87, 94, 91;

[0117] Substitute the parameters into the formula for calculation:

[0118] Calculate the numerator:

[0119]

[0120] Calculate the denominator:

[0121]

[0122] Calculate R 2 :

[0123]

[0124] The result of 0.914 showed that there was a positive correlation between texture characteristics and follicle maturity. The numerical value was used to help managers conduct artificial insemination and reproductive intervention at the appropriate time, improve the success rate of reproductive activities, optimize the breeding process, and improve economic benefits and sustainable development.

[0125] S403: Based on the follicle maturity analysis data, evaluate and record the maturity and size of follicles of multiple lactating sows, and generate follicle measurement data;

[0126] In sub-step S403, based on the follicle maturity analysis data, the maturity and size of the follicles of lactating sows are evaluated and recorded. The maturity stage of the follicles is determined by combining the follicle maturity analysis data and actual observations using biostatistical methods. The size measurement during the process is performed using digital image analysis technology. The target technology automatically identifies and measures the diameter of the follicles through image processing software. Data processing and analysis are completed using Excel and professional statistical software. The target software can provide fast and accurate data processing functions. The generated follicle measurement data records the maturity and specific size of each follicle, providing accurate data for formulating targeted breeding management measures.

[0127] See also Figure 6 Based on the follicle measurement data, combined with the feed consumption data, the effect of nutrient intake on follicle development is analyzed, and the influence of various physiological parameters of sows on follicle development is quantitatively evaluated. The steps to output the correlation analysis results are as follows:

[0128] S501: Based on the follicle measurement data, feed consumption records of multiple sows are collected, the nutritional intake of multiple sows is evaluated, and feed consumption data is generated;

[0129] In the above content, the feed consumption records of sows were collected based on the follicle measurement data to evaluate the nutritional intake of sows, and the total nutritional intake was calculated according to the formula I = f·E;

[0130] In the formula, I represents the total nutrient intake, f represents the nutrient content per unit of feed, and E represents the amount of feed consumed;

[0131] Detailed explanation of the formula and the process of formula calculation and derivation:

[0132] Assuming that the main component of feed is protein, the protein content per kilogram of feed is 200g, f = 0.2kg / kg, and assuming that the target sow consumes an average of 5kg of feed per day, E = 5kg / day, calculate I:

[0134] I = 0.2 × 5 = 1

[0135] The result I=1 indicates that the average daily nutrient intake of the sows during the recording period was 1 kg of protein. The calculation process is used to estimate the direct relationship between feed consumption and nutrient intake, and obtain the specific output of feed consumption data analysis. This quantitative nutrient intake assessment is crucial for the subsequent analysis of follicular development.

[0136] S502: Based on the feed consumption data, calculate the correlation between the nutrient intake and the follicle development parameters, including the speed of follicle development and the size of follicles, and generate a nutritional impact assessment result;

[0137] In sub-step S502, based on the feed consumption data, the statistical analysis software SPSS is used to perform data analysis, calculate the correlation between nutritional intake and follicular development parameters, analyze multiple variables including follicular development rate and follicular size, use correlation analysis and regression models to analyze the effects of different nutritional levels on follicular development, and use target analysis to determine how nutritional intake affects the growth of follicles. Pearson correlation analysis and multiple regression analysis are used to reveal the effects of different types and amounts of nutritional intake on follicular growth rate and size. The generated nutritional impact assessment results provide a scientific basis for adjusting feed formula and feeding strategies.

[0138] S503: Based on the results of the nutritional impact assessment, combined with the physiological parameters of multiple sows, including backfat thickness, body temperature, heart rate, and respiratory rate, identify and quantify the relationship between multiple physiological parameters and follicle development, and generate correlation analysis results;

[0139] In sub-step S503, based on the results of the nutritional impact assessment and combined with the physiological parameters of sows, including backfat thickness, body temperature, heart rate and respiratory rate, biostatistical methods and machine learning models are used to identify and quantify the relationship between the target physiological parameters and follicle development, including linear and nonlinear regression analysis. The target analysis is performed through the data processing software MATLAB and Python's SciPy library. Through target comprehensive analysis, how different physiological parameters interact with the developmental status of follicles is analyzed. The generated correlation analysis results provide in-depth insights for understanding the physiological changes of sows and provide data support for the implementation of more personalized breeding management strategies.

[0140] See also Figure 7 Based on the correlation analysis results, by identifying the time dependence in the follicle development pattern, a dynamic prediction model of follicle development was constructed to predict the follicle development status of multiple sows. The specific steps for generating follicle development prediction data are as follows:

[0141] S601: Based on the correlation analysis result, by performing time series analysis on the follicle development data, analyzing the periodic data changes of the follicle development state, and generating a time-dependent data set;

[0142] In sub-step S601, based on the results of correlation analysis, time series analysis techniques were applied to analyze the periodic changes in follicular development data. The autoregressive moving average (ARMA) model was used to capture the seasonal and trend components in the data. The statistical software R was used for implementation. The time series analysis process included a data stationarity test, Dickey-Fuller test to ensure that the data was suitable for model analysis, selection of appropriate lag terms and moving average terms, and optimization of model selection through the information criterion of model fitting. The generated time-dependent data set records the law of changes in follicular development status over time, providing accurate basic data for building a dynamic prediction model.

[0143] S602: Based on the time-dependent data set and the correlation between multiple physiological parameters of lactating sows and follicle development, a dynamic prediction model of follicle development is constructed, model parameters are adjusted to match actual observation data, and a prediction model adjustment record is generated;

[0144] In sub-step S602, a dynamic prediction model of follicular development is constructed based on the time-dependent data set. The multivariate time series analysis technology vector autoregression model is used to process the dynamic relationship of multiple time series data. In the model construction process, the adjustment of parameters is based on the principle of minimizing the prediction error. Data processing and model implementation are performed through the statistical software MATLAB and Python's pandas library to ensure that the model parameters best match the actual observed data, including multiple iterative adjustments to calibrate the model's prediction accuracy for the dynamics of follicular development. The generated prediction model adjustment record shows the parameter adjustment process, including the reasons and results of the adjustment, providing a transparent record for model verification and application.

[0145] S603: Based on the prediction model adjustment record, the model is verified by simulating multiple development scenarios, the follicle development status of multiple lactating sows is predicted, and follicle development prediction data is generated;

[0146] In sub-step S603, based on the prediction model adjustment record, the model is verified by simulating multiple development scenarios. Monte Carlo simulation technology is used to test the stability and prediction ability of the model by generating multiple development scenarios. The simulation is performed in the Python environment using numpy and scipy libraries, different initial conditions and input parameters are set, the model's predicted response to the state of follicular development is observed, and the performance of the model under different physiological parameter changes is evaluated to ensure that the model is used to provide accurate predictions within the range of real biological physiological changes. The generated follicular development prediction data includes a variety of model prediction results, demonstrating the wide applicability and prediction accuracy of the model, and providing a scientific basis for actual breeding decisions.

[0147] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for evaluating the degree of follicle development during sow lactation based on a neural network, characterized in that: The following steps are involved: Based on lactating sows, multiple sensors are used to collect the physiological parameters of sows during lactation in real time, measure and record body temperature, heart rate and respiratory rate data, and generate physiological parameter data sets; Based on the physiological parameter data set, the lactation cycle length data of multiple sows are recorded, and the change data of the sow weight and back fat thickness during the lactation cycle are monitored to generate sow lactation cycle data; Based on the sow lactation cycle data, ultrasonic images of lactating sows are collected, and follicle image morphology data is generated by segmenting the image area of ​​follicles, identifying follicle boundaries and structural features; Based on the follicle image morphological data, extract and analyze the multi-scale texture features of the follicle, evaluate the maturity and size of the follicle, and form follicle measurement data; Based on the follicle measurement data, combined with the feed consumption data, the effect of nutrient intake on follicle development is analyzed, and the effect of various physiological parameters of sows on follicle development is quantitatively evaluated, and the correlation analysis results are output; Based on the correlation analysis results, by identifying the time dependence in the follicular development pattern, a dynamic prediction model of follicular development is constructed to predict the follicular development status of multiple sows and generate follicular development prediction data.

2. The method for evaluating the degree of follicular development during sow lactation based on a neural network according to claim 1, characterized in that: The physiological parameter data set includes sow physiological parameter collection records, collection time information, and data preprocessing results. The sow lactation cycle data includes lactation cycle length, weight change data within the cycle, and back fat thickness change data within the cycle. The follicle image morphology data includes follicle image collection results, follicle boundary recognition data, and follicle structure feature information. The follicle measurement data includes multi-scale texture features of follicle images, follicle maturity scores, and follicle size measurement data. The correlation analysis results include the influence relationship between nutritional intake and follicular development, the quantitative evaluation results of the relationship between physiological parameters and follicular development, and the quantitative data of the influence of multiple variables on follicular development. The follicle development prediction data includes a dynamic prediction model, prediction results of follicular development status of multiple sows, and prediction model parameter optimization records.

3. The neural network-based method for evaluating the degree of follicular development during sow lactation according to claim 1, characterized in that: Based on lactating sows, a variety of sensors are used to collect the physiological parameters of sows during lactation in real time, measure and record body temperature, heart rate and respiratory rate data. The steps to generate the physiological parameter data set are as follows: Based on lactating sows, ultrasonic sensors, infrared sensors, heart rate, respiratory rate and body temperature sensors are used to collect and record the sow's back fat thickness, body temperature, heart rate and respiratory rate data in real time to generate physiological data collection results; Based on the physiological data collection results, formatting the collected physiological parameter data, including unit conversion and data structure standardization, to generate a standardized processing result; Based on the standardized processing results, the backfat thickness, body temperature, heart rate, and respiratory rate information are sorted according to the time information of data collection, outliers are identified and eliminated, data integrity is optimized, and a physiological parameter data set is generated.

4. The method for evaluating the degree of follicular development during sow lactation based on a neural network according to claim 1, characterized in that: Based on the physiological parameter data set, the lactation cycle length data of multiple sows are recorded, and the change data of the sow weight and back fat thickness during the lactation cycle are monitored. The steps of generating the sow lactation cycle data are specifically as follows: Based on the physiological parameter data set, by collecting the start time point and end time point information of the lactation cycles of multiple sows, recording the length of the lactation cycles of multiple sows, and generating lactation cycle record data; Based on the lactation cycle recorded data, by monitoring the changes in sow weight and backfat thickness at multiple stages in the lactation cycle, the weight and backfat thickness data are grouped according to the stages to generate weight and thickness change data; Based on the weight and thickness change data, the change pattern of backfat thickness during the lactation cycle is identified, the correlation between the cycle stage and backfat change is analyzed, and the sow lactation cycle data is generated.

5. The neural network-based method for evaluating the degree of follicular development during sow lactation according to claim 4, characterized in that: The specific formula for calculating the correlation between cycle phase and backfat change is: Among them, x i represents the lactation cycle phase of a single observation, represents the average value of the lactation cycle stage, y i represents the corresponding observed change in backfat thickness, represents the mean value of the change in backfat thickness, and r is the correlation coefficient between the two, which is used to quantify the strength and direction of the linear relationship between the lactation cycle stage and the change in backfat thickness.

6. The neural network-based method for evaluating the degree of follicular development during sow lactation according to claim 1, characterized in that: Based on the sow lactation cycle data, ultrasound images of lactating sows are collected, and the steps of segmenting the image area of ​​follicles, identifying the boundaries and structural features of follicles, and generating follicle image morphological data are specifically as follows: Based on the sow lactation cycle data, a lactating sow is scanned using an ultrasonic device to collect ultrasonic images of follicles and generate follicle image collection data; Based on the follicle image acquisition data, image preprocessing is performed, including adjusting brightness and contrast, optimizing image quality in combination with noise filtering, and generating a preprocessed follicle image set; Based on the pre-processed follicle image set, the follicle region in the image is identified and segmented, the boundary and structural features of the follicle are analyzed and marked, and the follicle image morphological data is generated.

7. The neural network-based method for evaluating the degree of follicular development during sow lactation according to claim 1, characterized in that: Based on the follicle image morphological data, the steps of extracting and analyzing the multi-scale texture features of the follicle, evaluating the maturity and size of the follicle, and forming the follicle measurement data are specifically as follows: Based on the follicle image morphology data, extracting texture features of the follicle in the image, including texture roughness, directionality and uniformity, to generate follicle texture feature data; Based on the follicle texture feature data, analyzing the correlation between texture parameters and follicle maturity, evaluating the correlation between texture features and follicle size, and generating follicle maturity analysis data; Based on the follicle maturity analysis data, the maturity and size of follicles of multiple lactating sows are evaluated and recorded to generate follicle measurement data.

8. The neural network-based method for evaluating the degree of follicular development during sow lactation according to claim 7, characterized in that: The specific formula for evaluating the correlation between texture features and follicle size is: Among them, y i represents the maturity score of the i-th follicle, f i represents the maturity of the i-th follicle predicted based on texture features, represents the average score of all follicles’ maturity, n represents the total number of follicle samples, and R 2 Represents the model explanation, which is used to measure the contribution of texture features to the prediction of follicle maturity.

9. The neural network-based method for evaluating the degree of follicular development during sow lactation according to claim 1, characterized in that: Based on the follicle measurement data, combined with the feed consumption data, the effect of nutrient intake on follicle development is analyzed, and the effect of various physiological parameters of sows on follicle development is quantitatively evaluated. The steps of outputting the correlation analysis results are specifically as follows: Based on the follicle measurement data, feed consumption records of multiple sows are collected, the nutritional intake of multiple sows is evaluated, and feed consumption data is generated; Based on the feed consumption data, calculating the correlation between nutrient intake and follicular development parameters, including the rate of follicular development and the size of follicles, and generating a nutritional impact assessment result; Based on the nutritional impact assessment results, combined with the physiological parameters of multiple sows, including backfat thickness, body temperature, heart rate, and respiratory rate, the relationship between multiple physiological parameters and follicular development is identified and quantified to generate correlation analysis results.

10. The neural network-based method for evaluating the degree of follicular development during sow lactation according to claim 1, characterized in that: Based on the correlation analysis results, by identifying the time dependency in the follicular development pattern, a dynamic prediction model for follicular development is constructed to predict the follicular development status of multiple sows. The steps of generating follicular development prediction data are specifically as follows: Based on the correlation analysis results, by performing time series analysis on the follicle development data, the periodic data changes of the follicle development state are analyzed to generate a time-dependent data set; Based on the time-dependent data set, according to the correlation between multiple physiological parameters of lactating sows and follicular development, a dynamic prediction model for follicular development is constructed, model parameters are adjusted to match actual observation data, and a prediction model adjustment record is generated; Based on the prediction model adjustment records, the model is verified by simulating multiple development scenarios, the follicular development status of multiple lactating sows is predicted, and follicular development prediction data is generated.

Images