Goose uniformity evaluation method and system based on artificial intelligence

Abstract

The invention discloses a goose uniformity evaluation method and system based on artificial intelligence. The method comprises the following steps: acquiring form data and behavior data of geese; generating a corresponding morphological feature vector according to the morphological data by using a pre-constructed deep morphological learning network; analyzing social interaction and behavior synchronism between geese by using a pre-constructed composite heterogeneous network model, and generating a corresponding behavior feature vector according to the behavior data; and evaluating the uniformity of the goose according to the morphological characteristic vector and the behavior characteristic vector. According to the embodiment of the invention, a more efficient and accurate artificial intelligence-driven goose uniformity evaluation scheme can be provided, so that the high-standard requirements of the modern breeding industry can be met.

Description

Technical Field

[0001] The invention belongs to the technical field of agricultural breeding, and in particular to an artificial intelligence-based goose uniformity evaluation method and system. Background Art

[0002] In animal husbandry, especially in the breeding industry of goose, geese are widely raised for their excellent meat quality and high production efficiency. In order to improve breeding efficiency and market competitiveness, the health status and uniformity of goose growth and development are issues of great concern to farmers. The overall uniformity of the goose flock is directly related to the breeding cost, meat quality and market value of the goose. Therefore, how to accurately and quickly evaluate the uniformity of geese has become an important task in the breeding industry.

[0003] Traditional uniformity assessment methods rely mainly on manual observation and individual sampling weighing to judge the overall growth status of a goose group through empirical judgment. This method is not only time-consuming and labor-intensive, but also highly subjective, and lacks accuracy and repeatability. In addition, it fails to provide information about the behavior patterns of the goose group, which often indicate changes in the health status of the group. In order to improve the accuracy and efficiency of the assessment, the breeding industry needs an objective, automated and intelligent solution. Summary of the invention

[0004] The purpose of the present invention is to provide a goose uniformity assessment method and system based on artificial intelligence to solve the deficiencies in the prior art and to provide a more efficient and accurate artificial intelligence-driven goose uniformity assessment solution to meet the high standards of modern breeding industry.

[0005] An embodiment of the present application provides a method for evaluating goose uniformity based on artificial intelligence, the method comprising:

[0006] Collect morphological and behavioral data of geese;

[0007] Using a pre-built deep morphological learning network, a corresponding morphological feature vector is generated according to the morphological data;

[0008] Using a pre-built composite heterogeneous network model to analyze social interactions and behavioral synchronization among geese, and generating corresponding behavioral feature vectors according to the behavioral data;

[0009] The uniformity of the goose is evaluated based on the morphological feature vector and the behavioral feature vector.

[0010] Optionally, the using a pre-built deep morphological learning network to generate a corresponding morphological feature vector according to the morphological data includes:

[0011] Construct and train a neural symbol conversion network, which is used to analyze and process morphological data by simulating the way the human nervous system processes information. The time-varying characteristics of neural symbols are used for deep learning training, and brain-like computing methods are used to adjust and optimize network parameters.

[0012] The morphological data is converted into a dynamic neural symbol sequence by using a trained neural symbol conversion network, and a morphological feature vector is extracted from the dynamic neural symbol sequence.

[0013] Optionally, the method of using a pre-built composite heterogeneous network model to analyze the social interaction and behavior synchronization among geese and generating a corresponding behavior feature vector according to the behavior data includes:

[0014] Construct and train a composite heterogeneous network model that combines a graph convolutional network and a time series analysis network, wherein the graph convolutional network is used to process structured graph information and the time series network is used to capture temporal dependencies in behavioral data;

[0015] The behavior data is input into the trained composite heterogeneous network model, and a corresponding behavior feature vector is output, wherein the feature vector comprehensively considers the individual behavior and group interaction behavior of the geese.

[0016] Optionally, the evaluating the uniformity of the goose according to the morphological feature vector and the behavioral feature vector includes:

[0017] According to the morphological feature vector and the behavioral feature vector, the behavior-morphological fusion index of the goose is calculated based on the following formula:

[0018]

[0019] Among them, the GUEF is the behavior-morphology fusion index, and the sigma q 2 is the variance of the behavioral data, the sigma m is the variance of the morphological data, the delta q is the behavior characteristic vector, the delta m is a morphological feature vector, theta and omega are dynamic adjustment weights;

[0020] The uniformity of the geese was assessed based on the behavior-morphology fusion index.

[0021] Optionally, the evaluating the uniformity of the goose according to the behavior-morphology fusion index includes:

[0022] Determine the numerical range of the behavior-morphology fusion index; wherein,

[0023] If it is greater than a first preset threshold, the uniformity of the goose is evaluated as a high level;

[0024] If it is greater than the second preset threshold and less than or equal to the first preset threshold, the uniformity of the goose is evaluated as medium;

[0025] If it is less than or equal to the second preset threshold, the uniformity of the goose is evaluated as a low level.

[0026] Another embodiment of the present application provides a goose uniformity assessment system based on artificial intelligence, the system comprising:

[0027] A collection module is used to collect morphological and behavioral data of geese;

[0028] A first generating module is used to generate a corresponding morphological feature vector according to the morphological data using a pre-built deep morphological learning network;

[0029] A second generation module is used to analyze the social interaction and behavior synchronization between geese by using a pre-built composite heterogeneous network model, and to generate a corresponding behavior feature vector according to the behavior data;

[0030] An evaluation module is used to evaluate the uniformity of the goose according to the morphological feature vector and the behavioral feature vector.

[0031] Yet another embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, wherein the computer program is configured to execute any of the above methods when running.

[0032] Yet another embodiment of the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute any of the methods described above.

[0033] Compared with the prior art, the present invention provides an artificial intelligence-based goose uniformity assessment method, which collects morphological data and behavioral data of the goose; uses a pre-constructed deep morphological learning network to generate corresponding morphological feature vectors according to the morphological data; uses a pre-constructed composite heterogeneous network model to analyze the social interaction and behavioral synchronization between geese, and generates corresponding behavioral feature vectors according to the behavioral data; and evaluates the uniformity of the goose based on the morphological feature vectors and the behavioral feature vectors, thereby being able to provide a more efficient and accurate artificial intelligence-driven goose uniformity assessment solution to meet the high standards of the modern breeding industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A hardware structure block diagram of a computer terminal for a goose uniformity evaluation method based on artificial intelligence provided by an embodiment of the present invention;

[0035] Figure 2 A schematic diagram of a process flow of a goose uniformity evaluation method based on artificial intelligence provided by an embodiment of the present invention;

[0036] Figure 3 A schematic diagram of the structure of a goose uniformity assessment system based on artificial intelligence provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, but should not be construed as limiting the present invention.

[0038] The embodiment of the present invention first provides a goose uniformity assessment method based on artificial intelligence, which can be applied to electronic devices such as computer terminals, specifically ordinary computers, etc.

[0039] The following describes it in detail by taking running on a computer terminal as an example. Figure 1 The hardware structure block diagram of a computer terminal for a goose uniformity evaluation method based on artificial intelligence provided by an embodiment of the present invention. Figure 1 As shown, the computer terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Optionally, the computer terminal may also include a transmission device 106 for communication functions and an input and output device 108. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0040] The memory 104 can be used to store software programs and modules of application software, such as program instructions / modules corresponding to the artificial intelligence-based goose uniformity evaluation method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0041] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of a computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0042] See also Figure 2 , an embodiment of the present invention provides a method for evaluating goose uniformity based on artificial intelligence, which may include the following steps:

[0043] S201, collecting morphological data and behavioral data of the goose; specifically, one implementation method may include:

[0044] Step 1: Biomarker and sensor deployment

[0045] - Biosimilar Sensor Development: Design and develop a new class of biosimilar sensors that mimic the physiological and behavioral characteristics of geese, such as body temperature, heart rate, and movement patterns. These sensors are designed to be nearly non-intrusive to the geese, who can be considered part of the flock.

[0046] -Sensor implantation: Sensors are implanted into selected geese through non-invasive means. Sensors are selectively implanted into individuals of different ages and genders to ensure representative and extensive data collection. The equipment should be able to capture details such as the goose's movement trajectory, interaction frequency, sound signals, and body changes.

[0047] Step 2: Multimodal Data Collection

[0048] -Deployment of multimodal observation equipment: Deploy multimodal observation equipment around the goose habitat, including high-resolution video cameras, infrared cameras, and sound monitoring equipment. These devices can simultaneously capture the morphological and behavioral data of the geese, providing information at both the visual and auditory levels. Capture the appearance data of the geese, such as feather color, size, and body shape, and also record the movement trajectory and group distribution status of the geese. Collect the behavioral data of the geese, such as movement speed, stay time, and frequency of interaction with other individuals.

[0049] -Synchronous observation and data recording: All observation equipment and sensors work synchronously to ensure the consistency of the collected data in time and the accurate correspondence between morphological and behavioral data. The collected raw data is preprocessed, including noise removal, filling missing values ​​and calibrating data. This step is to ensure data quality and lay a good foundation for subsequent analysis.

[0050] Through the above steps, the morphological and behavioral data of geese can be collected efficiently and accurately. The advantage of this method is that it can monitor and analyze in real time on a large scale, reduce human interference, and improve the quality of data and the accuracy of analysis.

[0051] S202, using a pre-built deep morphological learning network to generate a corresponding morphological feature vector according to the morphological data;

[0052] Specifically, a neural symbol conversion network can be constructed and trained as a deep morphological learning network, which is used to analyze and process morphological data by simulating the way the human nervous system processes information, wherein the time-varying characteristics of neural symbols are used for deep learning training, and brain-like computing methods are used to adjust and optimize network parameters;

[0053] Among them, the neural symbol conversion network is a neural network that aims to simulate the mechanism of information processing in the human brain and convert the input data into the form of neural symbols. This network usually takes into account the time-varying characteristics of the data and may contain biologically inspired neuron models and synaptic connections to better capture and simulate the dynamic processing capabilities of the human nervous system. In the goose uniformity assessment method, the neural symbol conversion network is used to parse the morphological data and produce complex and dynamic representations that capture the essence of the morphological characteristics of the goose. By simulating the processing method of the human brain, the network can extract rich features that are useful for the evaluation task while maintaining biological interpretability.

[0054] Neural symbols refer to the basic information units that flow and process in the neural symbol conversion network, which usually corresponds to neural pulses or electrical signals in biological neural networks. In the computational model, these symbols can be discrete pulse events or some form of continuous signals to represent the activity state of neurons. Neural symbols play the role of encoding and transmitting morphological data features in the model. By mapping morphological data into neural symbols, the neural symbol conversion network can process and analyze this information in a way close to the biological nervous system. The dynamic nature of neural symbols enables the network to adapt to changes in time, providing a refined time-space feature representation for subsequent uniformity evaluation.

[0055] Time-varying neural symbols can simulate the characteristics of how information changes over time in biological neural systems. Implementing this in neural networks means that the network can handle the temporal correlation of input data, such as dynamic morphological changes, which is critical for solving problems such as morphological recognition.

[0056] Brain-like computing technology simulates the structure and function of the human brain to improve machine learning algorithms. This includes but is not limited to neuromodulation, synaptic plasticity, and memory mechanisms. These mechanisms enable the network to not only learn adaptively, but also remain robust in the face of new, unknown, or changing data. One implementation method may include:

[0057] Step 1: Construction of Neural Symbolic Transformation Network

[0058] 1. Network structure design: Design a neural network model with multiple layers, each layer simulating a different neural processing level. Use biologically inspired neuron models, such as Spiking Neural Networks (SNN), which can process information using time-stamped pulses and are closer to the dynamics of real neural systems.

[0059] Each layer simulates a different neural processing mechanism. Neuron models in the network will adopt time-varying properties, such as dynamic synaptic strength and threshold variability, which can make the network response time-sensitive.

[0060] 2. Time-varying feature integration: integrating time-varying neurons, which have dynamic thresholds and plastic synaptic connections and can adjust their activation states according to the timing and intensity changes of input signals, so as to better handle dynamic changes in morphological data.

[0061] Integrate time-varying neurons in the neural network, whose activation functions depend not only on the current input but also on the previous state and input history. This dependence simulates the timeliness and memory characteristics of biological neural systems, which is particularly critical for processing and learning dynamic changes in morphological data.

[0062] Step 2: Application of brain-inspired computing methods

[0063] 1. Memory and learning mechanism: Implement memory-based learning mechanisms in the network, similar to human long-term memory and short-term memory. By adjusting the strength of synapses (called synaptic plasticity), the network can retain important information during training and forget unnecessary information.

[0064] Implementing neuromodulation mechanisms during network training simulates neurotransmitters such as dopamine or serotonin to adjust synaptic strength (synaptic plasticity). This adjustment allows the network to automatically strengthen or inhibit certain connections during learning, thereby optimizing response patterns and enhancing the effectiveness and efficiency of learning.

[0065] 2. Neuromodulation mechanism: Introducing neuromodulation mechanism, a bio-inspired technology that optimizes the learning process by simulating the effects of neurotransmitters such as dopamine. Neurotransmitters can regulate synaptic plasticity and enhance or inhibit the activation of specific neurons to adapt to different learning and decision-making environments.

[0066] A neural feedback mechanism is introduced to dynamically adjust the behavior of neurons based on the difference between the output and the expected result. This mechanism can adjust the parameters of the neural network in real time, ensuring that the network remains efficient and accurate when processing complex morphological data.

[0067] Step 3: Deep learning training and optimization

[0068] 1. Time-dependent training strategy: The time window method is used to perform time-dependent training on the neural network. In this training, the network not only learns morphological features, but also recognizes the patterns of these features changing over time.

[0069] A time-dependent training strategy is adopted to train the network to recognize and learn dynamic changes in morphological data by setting different time windows. This strategy ensures that the network can learn the time series characteristics in the data, such as gradual changes in morphology and periodic behavior.

[0070] 2. Optimization algorithm selection: Use optimization algorithms suitable for time-varying networks, such as gradient-based methods combined with back-propagation algorithms, which are specifically designed to handle the complexity of time series data. These algorithms can effectively adjust network parameters to improve prediction accuracy and processing power.

[0071] You can also use optimization algorithms with adaptive learning rates (such as Adam or RMSprop), which are particularly suitable for processing non-static and highly changing data sets. The algorithm automatically adjusts the learning rate according to the error changes during training, helping the network converge faster and improving the performance of the final model.

[0072] Step 4: Fine-grained tuning of network parameters

[0073] 1. Parameter fine-tuning: After the initial training is completed, fine-grained adjustments are made to further optimize the model’s performance. Use global optimization techniques such as genetic algorithms or simulated annealing to find the optimal network parameter settings.

[0074] In the later stages of network training, a meta-learning strategy can also be implemented to fine-tune the parameters, which can dynamically adjust the learning strategy according to the needs of different tasks. In addition, the network structure can be optimized and the risk of overfitting can be reduced through techniques such as Dropout or Batch Normalization.

[0075] 2. Performance evaluation and feedback: Regularly evaluate the performance of the network in processing morphological data, using metrics such as accuracy, recall, and F1 score. Adjust the learning rate, number of layers, number of neurons, etc. based on the evaluation results to ensure that the network can perform the task effectively. Regularly use the validation dataset to evaluate network performance and monitor metrics. Based on performance feedback, adjust the network structure (e.g., number of layers and number of neurons) and training parameters to achieve the best data processing effect.

[0076] Through this detailed and in-depth implementation, the Neural Symbol Transformation Network is not only able to simulate the processing methods of the human nervous system, but also can effectively analyze and process the morphological data of geese through brain-like computing methods and deep learning training of time-varying characteristics. This approach combines biological principles and artificial intelligence technology to solve practical problems.

[0077] The morphological data is converted into a dynamic neural symbol sequence by using a trained neural symbol conversion network, and a morphological feature vector is extracted from the dynamic neural symbol sequence.

[0078] Among them, the neural symbol sequence is a series of neural symbols arranged in chronological order, which reflects the activity of one or some neurons over time. This sequence can capture the dynamic patterns and temporal features in the input data, similar to the melody in music or the sentence structure in natural language. In the context of goose uniformity assessment, neural symbol sequences are used to represent the dynamic changes of morphological data over time. These sequences not only contain static information about the morphological characteristics of the goose, but also contain the dynamic relationship and change trend between the morphologies, thus providing more comprehensive information for evaluating the uniformity of the goose. Using this sequence data, the evaluation method can more accurately reflect the overall morphological characteristics of the goose and its consistency and changes in time, which is a key step in uniformity assessment.

[0079] The neural symbol sequences can be weighted to highlight the important neural symbols and suppress the unimportant ones, so as to emphasize the key features in the morphological data. The weighted neural symbol sequences are reduced in dimension to obtain a compressed high-level feature representation. The high-level features are further transformed linearly or nonlinearly to form a more comprehensive description of the morphological features. The feature vectors from different neural symbol sequences are spliced ​​or fused to generate a unified, multi-dimensional feature vector that takes into account the spatial properties and temporal changes of the morphology. Feature selection and dimensionality reduction can then be performed, such as through principal component analysis (PCA) or autoencoders to remove redundant information and retain the most representative features. The final feature vector is normalized to ensure that the features are on the same scale for subsequent statistical analysis and machine learning processing. The morphological feature vectors for goose uniformity assessment are output, which have encoded the appearance, morphological structural characteristics, and change trends over time of the goose.

[0080] The morphological data is converted into a dynamic neural symbol sequence, and then the morphological features are extracted from it to generate a high-dimensional morphological feature vector, each of which is a high-fidelity representation of the goose morphology. The benefits of this implementation are:

[0081] The way biological nervous systems process information is very efficient, especially when processing perceptual data. By emulating this mechanism, neural symbolic transformation techniques can process and parse complex morphological data more naturally and effectively. This approach can help the model capture subtle patterns and dynamic changes in the data, which may be overlooked in traditional static feature extraction methods.

[0082] Morphological features are not just static; in the real world, the morphological features of organisms will have subtle dynamic changes in different environments and conditions. Dynamic neural symbol sequences can capture such changes and provide a richer dimension of information, which is crucial for understanding the adaptability and variability of biological morphology.

[0083] Not only can the basic features of morphological data be extracted from the dynamic neural symbol sequence, but also the temporal dependencies in the data can be analyzed, which is particularly important for analyzing the ecological behavior and physiological changes of organisms such as geese. This method can extract deeper and implicit features, which are essential for high-precision classification, prediction or behavioral analysis.

[0084] S203, using a pre-built composite heterogeneous network model to analyze social interaction and behavioral synchronization among geese, and generating a corresponding behavioral feature vector according to the behavioral data;

[0085] Specifically, a composite heterogeneous network model can be constructed and trained, which combines a graph convolutional network and a time series analysis network, wherein the graph convolutional network is used to process structured graph information, and the time series network is used to capture the time dependency in the behavior data; the behavior data is input into the trained composite heterogeneous network model, and the corresponding behavior feature vector is output, wherein the feature vector comprehensively considers the individual behavior and group interaction behavior of geese. One implementation method may include:

[0086] Step 1: Graph Convolutional Network (GCN) design 1. Node feature extraction: A node is defined for each goose, and the node features include biological characteristics such as the goose's age, weight, and gender.

[0087] 2. Edge establishment and weight assignment: construct edges based on the frequency and density of interactions between geese, such as eating together, resting, and socializing. Use nonlinear functions such as Gaussian kernels to initialize edge weights, which can better reflect the intensity of different types of social interactions.

[0088] 3. Graph structure learning: Develop an adaptive graph structure learning mechanism to optimize the connections and weights of graph edges instead of defining them statically.

[0089] Step 2: Time Series Network (LSTM) Configuration 1. Feature sequence construction: Extract time-dependent features from behavioral data, such as activity level, feeding behavior, and temporal patterns of social behavior.

[0090] 2. Long short-term memory construction: Design a stacked LSTM network to process time series data, allowing behavioral patterns from short-term to long-term to be identified and learned.

[0091] Step 3: Fusion of heterogeneous networks 1. Feature space alignment: Align the output features of GCN and LSTM through the feature space to ensure that the two are fused in the same dimension.

[0092] 2. Dynamic fusion mechanism: Implement a dynamic weight fusion layer to dynamically adjust the fusion weights of GCN and LSTM output features by learning the contribution of different features to the prediction target.

[0093] 3. Introduction of attention mechanism: Integrate the attention mechanism in the fusion layer so that the model can autonomously learn to pay attention to behavioral features or social interaction patterns at certain key time points.

[0094] Step 4: End-to-end training and optimization 1. Multi-task learning framework: Set up an end-to-end framework that allows the model to simultaneously optimize the feature extraction tasks of social structure and behavioral temporal patterns.

[0095] 2. Loss function customization: Define the loss function to comprehensively consider the accurate representation of graph information and the capture of time-dependent features, such as combining graph regularization terms and sequence prediction errors.

[0096] 3. Back propagation and regularization: Use the back propagation algorithm to update the network weights, and introduce regularization terms to prevent overfitting.

[0097] Step 5: Behavior feature vector generation and interpretation 1. Feature vector extraction: In the final stage of the model, a feature extraction layer is set up, which aims to convert the high-level features extracted by GCN and time series network into a comprehensive behavior feature vector. This layer needs to be designed to maintain the important information of different source features and fuse them into a unified representation.

[0098] 2. Dimensionality reduction and feature selection: Apply dimensionality reduction techniques such as principal component analysis (PCA) or autoencoders to reduce the dimensionality of the fused features and select the most representative and discriminative features as the elements of the output feature vector.

[0099] 3. Feature vector normalization: The generated feature vector is normalized, for example using minimum-maximum scaling or z-score normalization methods, to ensure that the feature vector is comparable between features at different scales and suitable for subsequent analysis or modeling.

[0100] 4. Feature importance assessment: Introduce a feature importance assessment mechanism, such as analysis methods based on model weights or feature contributions, to help understand which features play an important role in the behavioral feature vector and provide guidance for subsequent research.

[0101] 5. Enhanced model interpretability: Although the extracted feature vector itself is mainly used for further analysis, in order to improve the interpretability of the model, techniques such as LI ME (Local Interpretable Model-Opaque Explanations) or SHAP (SHapley Added Explanations) can be introduced to provide more explanations about the model output decisions.

[0102] 6. Output layer construction: Ensure that the output layer can generate behavioral feature vectors in the expected format, which can be used for subsequent processing and can also be used to monitor the welfare of geese or predict potential behavioral changes.

[0103] Through the above steps, the composite heterogeneous network model can extract key information for in-depth analysis of goose social behavior while maintaining the practicality and interpretability of the output behavior feature vector. The implementation of this method aims to generate a comprehensive and accurate description of behavioral characteristics through multi-dimensional feature fusion, providing a powerful tool for goose behavior research and breeding management.

[0104] S204: Evaluate the uniformity of the goose according to the morphological feature vector and the behavioral feature vector.

[0105] Specifically, the behavior-morphology fusion index of the goose can be calculated according to the morphological feature vector and the behavioral feature vector based on the following formula:

[0106]

[0107] Among them, GUEF is the behavior-morphology fusion index, which is a comprehensive measure of goose uniformity, taking into account both behavioral data and morphological data; sigma q 2 is the variance of the behavioral data, sigma m is the variance of the morphological data, which measures the differences and fluctuations in goose behavior. The larger the variance, the greater the behavioral differences between individuals. q is a behavioral feature vector, generated by a composite heterogeneous network model based on behavioral data, reflecting the individual behavioral characteristics of the goose; delta m is a morphological feature vector, generated by a deep morphological learning network based on morphological data, reflecting the morphological characteristics of the goose; theta and omega are dynamically adjusted weights used to balance the influence of morphological data and behavioral data in uniformity assessment, allowing the assessment process to adjust the importance of these two types of data according to different situations. The weights theta and omega can be adjusted to reflect different assessment needs and conditions, such as at a specific time, morphological differences can better reflect uniformity than behavioral differences, or vice versa.

[0108] The design concept of this indicator is based on the concept of variance in statistics, which is believed to measure the similarity between individuals in a group. The smaller the variance, the higher the uniformity of the group. The logarithmic transformation of the feature vector helps to emphasize the relative importance of each element in the feature vector while reducing the impact of outliers. By summing the negative exponents of the two variances, a comprehensive score of group behavior and morphological differences can be obtained, which reflects the overall uniformity of the goose group.

[0109] The uniformity of the geese was assessed based on the behavior-morphology fusion index.

[0110] Specifically, the value range of the behavior-morphology fusion index may be determined;

[0111] First, the GUEF value needs to be calculated and analyzed. The GUEF value is a comprehensive indicator that combines the behavioral data and morphological data of the goose, which can reflect the overall uniformity of the goose. This value is calculated by the formula described in claim 4. This step involves comparing the GUEF value with a pre-set threshold to classify the uniformity of the goose.

[0112] Two key thresholds need to be set: the first preset threshold and the second preset threshold. These thresholds can be set based on historical data analysis, statistical methods or experience. They represent the boundaries of the geese uniformity classification. The first preset threshold is used to distinguish between high uniformity and medium uniformity, while the second preset threshold is used to distinguish between medium uniformity and low uniformity. Based on the comparison of the GUEF value with the preset thresholds, the uniformity of the geese is divided into three levels: high, medium and low. This step is the core of the evaluation process, which converts quantitative data into classification results that are easy to understand and use.

[0113] Among them, if it is greater than the first preset threshold, the uniformity of the goose is evaluated as a high level; if it is greater than the second preset threshold and less than or equal to the first preset threshold, the uniformity of the goose is evaluated as a medium level; if it is less than or equal to the second preset threshold, the uniformity of the goose is evaluated as a low level.

[0114] If the GUEF value is greater than a first preset threshold, the uniformity of the geese is considered to be high, indicating that the morphological and behavioral characteristics of the geese are very consistent in the group.

[0115] If the GUEF value is greater than the second preset threshold and less than or equal to the first preset threshold, the uniformity of the geese is evaluated as medium, which means that there are some differences in the morphological and behavioral characteristics of the geese in the group, but they are still within an acceptable range.

[0116] If the GUEF value is less than or equal to the second preset threshold, the uniformity of the geese is assessed to be low, which may mean that there are significant individual differences in the group and management measures may be needed to improve uniformity.

[0117] The evaluation results can be applied to goose breeding management. For example, a low-uniformity group may need to optimize the feeding environment or feed formula, while a high-uniformity group may mean that the current breeding strategy is successful. The evaluation results can also be used as a basis for selecting breeding individuals, thereby improving the uniformity of the goose population at the genetic level.

[0118] The uniformity assessment method can be iterated and optimized based on the feedback obtained from the actual application. It may be necessary to adjust the threshold value according to new data or changed environmental conditions, or improve the calculation method of GUEF value. In addition, the accuracy of the entire assessment process can be verified and improved by continuously monitoring the results of field application.

[0119] It can be seen that by collecting the morphological data and behavioral data of the goose; using the pre-constructed deep morphological learning network to generate the corresponding morphological feature vector according to the morphological data; using the pre-constructed composite heterogeneous network model to analyze the social interaction and behavioral synchronization between the geese, and generating the corresponding behavioral feature vector according to the behavioral data; and evaluating the uniformity of the geese according to the morphological feature vectors and the behavioral feature vectors, it is possible to provide a more efficient and accurate artificial intelligence-driven goose uniformity evaluation solution to meet the high standards of the modern breeding industry.

[0120] Another embodiment of the present invention provides a goose uniformity evaluation system based on artificial intelligence, see Figure 3 , the system may include:

[0121] A collection module 301 is used to collect morphological data and behavioral data of the goose;

[0122] A first generating module 302 is used to generate a corresponding morphological feature vector according to the morphological data using a pre-built deep morphological learning network;

[0123] The second generation module 303 is used to use the pre-built composite heterogeneous network model to analyze the social interaction and behavior synchronization between geese, and generate corresponding behavior feature vectors according to the behavior data;

[0124] The evaluation module 304 is used to evaluate the uniformity of the goose according to the morphological feature vector and the behavioral feature vector.

[0125] It can be seen that by collecting the morphological data and behavioral data of the goose; using the pre-constructed deep morphological learning network to generate the corresponding morphological feature vector according to the morphological data; using the pre-constructed composite heterogeneous network model to analyze the social interaction and behavioral synchronization between the geese, and generating the corresponding behavioral feature vector according to the behavioral data; and evaluating the uniformity of the geese according to the morphological feature vectors and the behavioral feature vectors, it is possible to provide a more efficient and accurate artificial intelligence-driven goose uniformity evaluation solution to meet the high standards of the modern breeding industry.

[0126] An embodiment of the present invention further provides a storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.

[0127] Specifically, in this embodiment, the above storage medium may be configured to store a computer program for performing the following steps:

[0128] S201, collecting morphological data and behavioral data of the goose;

[0129] S202, using a pre-built deep morphological learning network to generate a corresponding morphological feature vector according to the morphological data;

[0130] S203, using a pre-built composite heterogeneous network model to analyze social interaction and behavioral synchronization among geese, and generating a corresponding behavioral feature vector according to the behavioral data;

[0131] S204: Evaluate the uniformity of the goose according to the morphological feature vector and the behavioral feature vector.

[0132] It can be seen that by collecting the morphological data and behavioral data of the goose; using the pre-constructed deep morphological learning network to generate the corresponding morphological feature vector according to the morphological data; using the pre-constructed composite heterogeneous network model to analyze the social interaction and behavioral synchronization between the geese, and generating the corresponding behavioral feature vector according to the behavioral data; and evaluating the uniformity of the geese according to the morphological feature vectors and the behavioral feature vectors, it is possible to provide a more efficient and accurate artificial intelligence-driven goose uniformity evaluation solution to meet the high standards of the modern breeding industry.

[0133] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0134] Specifically, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0135] Specifically, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0136] S201, collecting morphological data and behavioral data of the goose;

[0137] S202, using a pre-built deep morphological learning network to generate a corresponding morphological feature vector according to the morphological data;

[0138] S203, using a pre-built composite heterogeneous network model to analyze social interaction and behavioral synchronization among geese, and generating a corresponding behavioral feature vector according to the behavioral data;

[0139] S204: Evaluate the uniformity of the goose according to the morphological feature vector and the behavioral feature vector.

[0140] Specifically, the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0141] It can be seen that by collecting the morphological data and behavioral data of the goose; using the pre-constructed deep morphological learning network to generate the corresponding morphological feature vector according to the morphological data; using the pre-constructed composite heterogeneous network model to analyze the social interaction and behavioral synchronization between the geese, and generating the corresponding behavioral feature vector according to the behavioral data; and evaluating the uniformity of the geese according to the morphological feature vectors and the behavioral feature vectors, it is possible to provide a more efficient and accurate artificial intelligence-driven goose uniformity evaluation solution to meet the high standards of the modern breeding industry.

[0142] The above describes in detail the structure, features and effects of the present invention based on the embodiments shown in the drawings. The above is only a preferred embodiment of the present invention, but the present invention is not limited to the scope of implementation shown in the drawings. Any changes made according to the concept of the present invention, or modifications to equivalent embodiments with equivalent changes, which still do not exceed the spirit covered by the description and drawings, should be within the protection scope of the present invention.

Claims

1. A method for evaluating goose uniformity based on artificial intelligence, characterized in that: The method comprises: Collect morphological and behavioral data of geese; Using a pre-built deep morphological learning network, a corresponding morphological feature vector is generated according to the morphological data; Using a pre-built composite heterogeneous network model to analyze social interactions and behavioral synchronization among geese, and generating corresponding behavioral feature vectors according to the behavioral data; The uniformity of the goose is evaluated based on the morphological feature vector and the behavioral feature vector.

2. The method according to claim 1, characterized in that The method of using a pre-built deep morphological learning network to generate a corresponding morphological feature vector according to the morphological data includes: Construct and train a neural symbol conversion network, which is used to analyze and process morphological data by simulating the way the human nervous system processes information. The time-varying characteristics of neural symbols are used for deep learning training, and brain-like computing methods are used to adjust and optimize network parameters. The morphological data is converted into a dynamic neural symbol sequence by using a trained neural symbol conversion network, and a morphological feature vector is extracted from the dynamic neural symbol sequence.

3. The method according to claim 2, characterized in that The method of using a pre-built composite heterogeneous network model to analyze social interaction and behavioral synchronization among geese and generating corresponding behavioral feature vectors according to the behavioral data includes: Construct and train a composite heterogeneous network model that combines a graph convolutional network and a time series analysis network, wherein the graph convolutional network is used to process structured graph information and the time series network is used to capture temporal dependencies in behavioral data; The behavior data is input into the trained composite heterogeneous network model, and a corresponding behavior feature vector is output, wherein the feature vector comprehensively considers the individual behavior and group interaction behavior of the geese.

4. The method according to claim 3, characterized in that: The step of evaluating the uniformity of the goose according to the morphological feature vector and the behavioral feature vector comprises: According to the morphological feature vector and the behavioral feature vector, the behavior-morphological fusion index of the goose is calculated based on the following formula: Among them, the GUEF is the behavior-morphology fusion index, and the sigma q 2 is the variance of the behavioral data, the sigma m is the variance of the morphological data, the delta q is the behavior characteristic vector, the delta m is a morphological feature vector, theta and omega are dynamic adjustment weights; The uniformity of the geese was assessed based on the behavior-morphology fusion index.

5. The method according to claim 4, characterized in that The step of evaluating the uniformity of geese according to the behavior-morphology fusion index includes: Determine the numerical range of the behavior-morphology fusion index; wherein, If it is greater than a first preset threshold, the uniformity of the goose is evaluated as a high level; If it is greater than the second preset threshold and less than or equal to the first preset threshold, the uniformity of the goose is evaluated as medium; If it is less than or equal to the second preset threshold, the uniformity of the goose is evaluated as a low level.

6. An artificial intelligence-based goose uniformity assessment system, characterized in that: The system comprises: A collection module is used to collect morphological and behavioral data of geese; A first generating module is used to generate a corresponding morphological feature vector according to the morphological data using a pre-built deep morphological learning network; A second generation module is used to analyze the social interaction and behavior synchronization between geese by using a pre-built composite heterogeneous network model, and to generate a corresponding behavior feature vector according to the behavior data; An evaluation module is used to evaluate the uniformity of the goose according to the morphological feature vector and the behavioral feature vector.

7. The system according to claim 6, characterized in that The first generating module is specifically used for: Construct and train a neural symbol conversion network, which is used to analyze and process morphological data by simulating the way the human nervous system processes information. The time-varying characteristics of neural symbols are used for deep learning training, and brain-like computing methods are used to adjust and optimize network parameters. The morphological data is converted into a dynamic neural symbol sequence by using a trained neural symbol conversion network, and a morphological feature vector is extracted from the dynamic neural symbol sequence.

8. The system according to claim 7, characterized in that The second generating module is specifically used for: Construct and train a composite heterogeneous network model that combines a graph convolutional network and a time series analysis network, wherein the graph convolutional network is used to process structured graph information and the time series network is used to capture temporal dependencies in behavioral data; The behavior data is input into the trained composite heterogeneous network model, and a corresponding behavior feature vector is output, wherein the feature vector comprehensively considers the individual behavior and group interaction behavior of the geese.

9. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 5 when executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 5.

Images