A beef cattle breeding monitoring system and method
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 丰都县动物疫病预防控制中心
- Filing Date
- 2026-02-05
- Publication Date
- 2026-06-09
Smart Images
Figure CN122176788A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of visual monitoring technology, and more specifically, to a monitoring system and method for beef cattle farming. Background Technology
[0002] Visual monitoring in beef cattle farming is the process of collecting and analyzing the behavior, physiological state, and group dynamics of individual beef cattle in real time using computer vision technology. This helps farms achieve intelligent health management and precision farming. Its goal is to obtain the health performance, nutritional status, and behavioral patterns of individual beef cattle and groups at different growth stages, thereby providing reliable data support for precision feeding, disease early warning, and control of the farming environment, and improving the management efficiency and farming benefits of modern large-scale farms.
[0003] However, current technologies for monitoring beef cattle farming typically rely on manual inspections or coarse-grained monitoring methods based on single sensors, such as scales or activity counters. These methods struggle to comprehensively capture the multi-dimensional characteristics of beef cattle, including posture, gait, body contours, and relative position within the herd. They also fail to effectively extract continuous movement patterns at different behavioral stages, resulting in coarse and delayed identification of health trends. Furthermore, the lack of correlation analysis with environmental factors makes it difficult to detect risks such as malnutrition, abnormal movement, disease precursors, or stress responses in a timely manner, thus affecting the accurate diagnosis and timely intervention of abnormal conditions. Therefore, achieving fine-grained visual monitoring of beef cattle health status in complex farming environments has become a significant challenge for the industry. Summary of the Invention
[0004] This application provides a beef cattle farming monitoring system and method, which can realize fine-grained visual monitoring of the health status of beef cattle in complex farming environments.
[0005] Firstly, this application provides a visual monitoring method for the status of beef cattle farming, applied to a beef cattle farming monitoring system. The method includes the following steps: Real-time image sequences of individual beef cattle in the beef cattle breeding area are collected, and a multi-dimensional visual representation of the beef cattle target is constructed based on the real-time image sequences; The action pattern features of beef cattle at different behavioral stages are extracted from the multidimensional visual representation, and the dynamic trajectory of beef cattle behavior in the beef cattle breeding area is generated based on the action pattern features at different behavioral stages. Based on the aforementioned behavioral dynamic trajectory, the health change trend of beef cattle in different breeding periods is identified in a fine-grained segmentation, and the local feature vector of the health status of beef cattle in each breeding period is obtained. The behavioral stability of health monitoring of beef cattle in different breeding periods is determined by all the local feature vectors. Based on the correlation changes between the behavioral stability and the environmental data of the beef cattle population, a visual risk index of the health status of beef cattle in the beef cattle breeding area is determined; Based on the visualized risk index, monitoring results are generated for livestock management in beef cattle breeding areas.
[0006] In conjunction with the first aspect, in one possible implementation, constructing a multidimensional visual representation of the beef cattle target based on the real-time image sequence specifically includes: Individual cattle detection and segmentation are performed on the real-time image sequence; Based on the segmentation results, the posture key points, gait motion vectors and body surface contour features of individual beef cattle were extracted. Determine the relative position matrix among individual beef cattle within the herd; The posture key points, gait motion vectors, body surface contour features, and relative position matrix are fused to generate a multidimensional visual representation of the beef cattle target.
[0007] In conjunction with the first aspect, in one possible implementation, extracting the action pattern features of beef cattle at different behavioral stages from the multidimensional visual representation specifically includes: Behavioral stage identification is performed on the multidimensional visual representation to obtain behavioral stage labels for beef cattle, including feeding, movement, and rest. Extract stage-specific action features based on labels for each behavioral stage; All stage-specific action characteristics were organized chronologically into action pattern characteristics of beef cattle at different behavioral stages.
[0008] In conjunction with the first aspect, in one possible implementation, generating the dynamic trajectory of beef cattle behavior in a beef cattle farming area based on the action pattern characteristics of the different behavioral stages specifically includes: Based on the aforementioned action pattern features, an initial behavioral state sequence of beef cattle in the beef cattle breeding area is constructed; The initial behavior state sequence is subjected to time-series smoothing to generate an optimized state sequence; The dynamic trajectory of beef cattle behavior in the beef cattle breeding area is generated based on the optimized state sequence.
[0009] In conjunction with the first aspect, in one possible implementation, fine-grained segmentation identification of the health change trends of beef cattle at different breeding periods is performed based on the aforementioned behavioral dynamic trajectory to obtain local feature vectors of the health status of beef cattle at each breeding period. Specifically, this includes: The dynamic trajectory of the behavior is segmented into fine-grained segments based on the preset breeding time period to obtain the trajectory segments corresponding to each breeding time period; Statistical and frequency domain features characterizing the health trends of beef cattle are extracted from trajectory segments of each breeding period. The extracted statistical features and frequency domain features are fused into a local feature vector of the health status of beef cattle in the corresponding breeding period.
[0010] In conjunction with the first aspect, one possible implementation involves determining the behavioral stability of beef cattle health monitoring at different rearing stages using all local feature vectors, specifically including: All local feature vectors are organized into a temporal feature sequence in chronological order; Determine the similarity matrix between adjacent feature vectors in the time-series feature sequence; The similarity matrix was used to determine the behavioral stability of beef cattle health monitoring at different breeding periods.
[0011] In conjunction with the first aspect, in one possible implementation, determining the visualized risk index of beef cattle health status in a beef cattle farming area based on the correlation changes between the behavioral stability and the environmental data of the beef cattle population specifically includes: Obtain environmental data for beef cattle herds; Determine the correlation and change patterns between the behavioral stability and environmental data; Based on the aforementioned correlation change patterns, a visualized risk index is determined for the health status of beef cattle in beef cattle farming areas, including those with malnutrition, abnormal movement, suspected diseases, and stress responses.
[0012] Secondly, this application provides a beef cattle farming monitoring system, including a visual monitoring unit, wherein the visual monitoring unit includes: The acquisition module is used to acquire real-time image sequences of individual beef cattle in the beef cattle breeding area, and to construct a multi-dimensional visual representation of the beef cattle target based on the real-time image sequences. The processing module is used to extract the action pattern features of beef cattle at different behavioral stages from the multidimensional visual representation, and generate the dynamic trajectory of beef cattle behavior in the beef cattle breeding area based on the action pattern features at different behavioral stages. The processing module is also used to perform fine-grained segmentation identification of the health change trend of beef cattle in different breeding periods based on the behavioral dynamic trajectory, obtain local feature vectors of the health status of beef cattle in each breeding period, and determine the behavioral stability of health monitoring of beef cattle in different breeding periods through all local feature vectors. The processing module is also used to determine a visual risk index of the health status of beef cattle in the beef cattle breeding area based on the correlation changes between the behavioral stability and the environmental data of the beef cattle population. The execution module is used to generate monitoring results for livestock management in beef cattle breeding areas based on the visualized risk index.
[0013] Thirdly, this application provides a computer device, which includes a memory and a processor. The memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the computer device performs the above-described visual monitoring method for the state of beef cattle farming.
[0014] Fourthly, this application provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the aforementioned visual monitoring method for the state of beef cattle farming.
[0015] The technical solution provided in this application has the following beneficial effects: This application's solution, firstly, generates dynamic behavioral trajectories of beef cattle in a beef cattle farming area based on the action pattern characteristics of different behavioral stages. This allows for the reconstruction of the beef cattle's behavioral evolution process in complex farming environments, avoiding information loss caused by single-frame images and more accurately capturing the correlation between abnormal behaviors, transforming health changes from rough judgment to process-level monitoring. Secondly, based on the dynamic behavioral trajectories, fine-grained segmentation and identification of the health change trends of beef cattle at different farming periods are performed, obtaining local feature vectors of the beef cattle's health status at each farming period. By using all local feature vectors, the behavioral stability of beef cattle health monitoring at different farming periods is determined. This divides long-term behavioral sequences into local periods with specific growth or activity characteristics, meticulously representing minute health fluctuations. Furthermore, it can identify whether behavioral changes are continuous and stable, or whether sudden fluctuations related to health problems occur, thereby ensuring... The health trend assessment is robust and reliable. Then, based on the correlation changes between the behavioral stability and the environmental data of the beef cattle population, a visualized risk index for the health status of beef cattle in the beef cattle farming area is determined. This index can identify the causal or induced relationship between behavioral abnormalities and environmental changes, quantifying complex behavior-environment coupling patterns into intuitive risk symbols. This allows ranch managers to quickly understand the true health status of beef cattle in complex environments, achieving a leap from feature analysis to risk identification in fine-grained monitoring. Finally, monitoring results for livestock management in beef cattle farming areas are generated based on the visualized risk index. This allows ranch managers to quickly locate abnormal individuals without manually judging a large number of images, thereby improving the fine-grainedness, timeliness, and effectiveness of beef cattle health management and achieving intelligent health monitoring in complex scenarios. In summary, this solution can achieve fine-grained visual monitoring of beef cattle health status in complex farming environments. Attached Figure Description
[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is an exemplary flowchart of a visual monitoring method for beef cattle farming status according to some embodiments of this application; Figure 2 This is an exemplary flowchart illustrating the extraction of action pattern features according to some embodiments of this application; Figure 3 This is an exemplary flowchart illustrating fine-grained segmentation recognition according to some embodiments of this application; Figure 4 This is a schematic diagram of the structure of a visual monitoring unit according to some embodiments of this application; Figure 5 This is a schematic diagram of the structure of a computer device for implementing a visual monitoring method for beef cattle farming status according to some embodiments of this application. Detailed Implementation
[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0019] refer to Figure 1 The figure is an exemplary flowchart of a visual monitoring method for the state of beef cattle farming according to some embodiments of this application. The visual monitoring method for the state of beef cattle farming mainly includes the following steps: In step 101, real-time image sequences of individual beef cattle in the beef cattle breeding area are collected, and a multi-dimensional visual representation of the beef cattle target is constructed based on the real-time image sequences.
[0020] In specific implementation, the real-time image sequence of individual beef cattle in the beef cattle breeding area can be achieved in the following way: multiple high-definition network cameras deployed in the beef cattle breeding area can continuously capture video streams at a set frame rate. The cameras can be installed in locations such as the ceiling, fence side walls, etc., to cover the main areas of beef cattle activity, such as feeding areas, rest areas, and exercise areas, and wide-angle lenses are used to ensure no blind spots in the field of view. The frame rate is one frame per second. The cameras are connected to a central processing server via Ethernet or wireless network. The server decodes the video stream and divides it into discrete image frames in sequence, thereby generating a real-time image sequence of individual beef cattle in the beef cattle breeding area. At the same time, to adapt to different lighting conditions, such as day and night changes, some cameras can integrate infrared night vision functions and automatically perform white balance and exposure adjustments during the acquisition process to ensure that the image quality meets the requirements of subsequent processing. In addition, the acquisition system can be configured with timed tasks or manual triggers to ensure the continuity and integrity of the image sequence. Other methods can also be used for acquisition in other embodiments, which are not specifically limited here.
[0021] It should be noted that the real-time image sequence in this application refers to a set of visual data describing the dynamic behavior and state of individual beef cattle.
[0022] In some embodiments, constructing a multidimensional visual representation of the beef cattle target based on the real-time image sequence can be achieved through the following steps: Individual cattle detection and segmentation are performed on the real-time image sequence; Based on the segmentation results, the posture key points, gait motion vectors and body surface contour features of individual beef cattle were extracted. Determine the relative position matrix among individual beef cattle within the herd; The posture key points, gait motion vectors, body surface contour features, and relative position matrix are fused to generate a multidimensional visual representation of the beef cattle target.
[0023] In specific implementation, the detection and segmentation of individual cattle in the real-time image sequence can be achieved in the following way: First, each frame of the real-time image sequence can be input into a pre-trained target detection model, such as the YOLOv5 network, to locate the individual cattle. The model outputs the bounding box coordinates of each cattle target. Then, the image region containing the bounding box is input into a semantic segmentation model, such as the DeepLabv3+ network. The model classifies each pixel and outputs a binary segmentation mask that depicts the body range of the cattle. Finally, the bounding box coordinates and the segmentation mask are used together as the detection and segmentation results.
[0024] In specific implementation, the extraction of posture key points, gait motion vectors, and body surface contour features of individual beef cattle based on the segmentation results can be achieved in the following way: First, based on the beef cattle body region defined by the segmentation mask in the segmentation results, a posture estimation model, such as the OpenPose network, can be used to identify and output the two-dimensional coordinate sequence of key joints of the beef cattle's trunk and limbs as the posture key points of the individual beef cattle; then, continuous frames are extracted from the real-time image sequence, and optical flow methods, such as the Farneback algorithm, are applied within the segmentation mask region to calculate the motion vectors of pixels, and these vectors are statistically analyzed to characterize the gait rhythm and amplitude, forming gait motion vectors; simultaneously, the contour of the segmentation mask is extracted, and its geometric features such as Hu moment, area, and perimeter ratio are calculated as body surface contour features; other methods can also be used in other embodiments, which are not limited here.
[0025] In practice, the relative position matrix between individual beef cattle within a herd can be determined as follows: Based on the bounding box coordinates of all beef cattle targets in the segmentation results, calculate the coordinates of the center point of the bounding box of each individual beef cattle; then, based on the coordinates of all center points, calculate the Euclidean distance between any two center points; finally, arrange these distance values in a fixed order of individual beef cattle to construct an N×N symmetric matrix, where N is the number of individuals. Each element (i, j) of this matrix represents the spatial relative distance between individual beef cattle i and individual beef cattle j. This matrix is the relative position matrix between individual beef cattle within a herd.
[0026] In specific implementation, the multidimensional visual representation of the beef cattle target is generated by fusing the posture key points, gait motion vectors, body surface contour features, and relative position matrix. This can be achieved in the following way: First, the posture key points, gait motion vectors, and body surface contour features are flattened and normalized respectively, converting them into multiple one-dimensional feature vectors. Next, the relative position matrix is also flattened, converting it into a one-dimensional vector. Then, all the above one-dimensional vectors are concatenated to form a comprehensive global feature vector. Finally, this global feature vector is standardized again and output as a multidimensional visual representation of the beef cattle target in the beef cattle breeding area, which fully encapsulates the posture, gait, body surface, and spatial position information of a single beef cattle in the herd. Other methods can also be used in other embodiments, which are not limited here.
[0027] It should be noted that the segmentation results in this application refer to the basic data used to identify the pixel range of individual beef cattle and their positions in the image; the pose key points in this application refer to the structured data used to quantitatively describe the instantaneous posture and orientation of the beef cattle's body in space; the gait motion vector in this application refers to the dynamic features used to quantitatively describe the rhythm, amplitude, and coordination of the beef cattle's limb movements; the body surface contour features in this application refer to the morphological parameters used to quantitatively describe the external shape and fullness of the beef cattle; the relative position matrix in this application refers to the structured data used to describe the instantaneous spatial distance relationship between individuals within the beef cattle herd; and the multidimensional visual representation in this application refers to a comprehensive feature vector that integrates information on the posture, behavior, body condition, and spatial relationships of individual beef cattle in the beef cattle breeding area.
[0028] In step 102, the action pattern features of beef cattle at different behavioral stages are extracted from the multidimensional visual representation, and the dynamic trajectory of beef cattle behavior in the beef cattle breeding area is generated based on the action pattern features of different behavioral stages.
[0029] In some embodiments, reference Figure 2 As shown, this figure is an exemplary flowchart of extracting action pattern features in some embodiments of this application. In this embodiment, extracting action pattern features of beef cattle at different behavioral stages from the multidimensional visual representation can be achieved by the following steps: First, in step 1021, the multidimensional visual representation is used to identify behavioral stages to obtain behavioral stage labels for beef cattle, including feeding, exercise, and rest. Secondly, in step 1022, stage-specific action features are extracted based on the labels of each behavioral stage; Finally, in step 1023, all stage-specific action features are organized in chronological order into action pattern features of beef cattle at different behavioral stages.
[0030] In specific implementation, the behavioral stage identification of the multidimensional visual representation to obtain behavioral stage labels for beef cattle including feeding, movement, and rest can be achieved in the following way: the multidimensional visual representation arranged in chronological order can be input into a pre-trained long short-term memory network classification model. This model uses the hidden state of the previous time step combined with the multidimensional visual representation vector of the current time step as input, and outputs the behavioral stage probability distribution corresponding to each time step through a fully connected layer and a softmax activation function. The category corresponding to the maximum probability is taken as the behavioral stage label at that time step, including feeding, movement, and rest. The long short-term memory network classification model can be trained through a labeled beef cattle behavior dataset and can identify the transition of behavioral patterns based on temporal features such as posture key point sequences and gait motion vectors. Other methods can also be used in other embodiments, which are not limited here.
[0031] In specific implementation, the extraction of stage-specific action features based on the labels of each behavioral stage can be achieved in the following way: the multidimensional visual representation can be divided into segments of different behavioral stages of beef cattle according to the labels of each behavioral stage. For example, for the feeding stage, the frequency of head pitch angle change and the movement amplitude of the mouth near the ground area can be extracted; for the movement stage, the variance of the swing angle of the key points of the limb joints and the average amplitude of the gait movement vector can be extracted; for the resting stage, the spatial position stability index of the key points of the trunk can be extracted. All of these features can be statistically calculated through a sliding window and normalized into a fixed-dimensional feature vector to form a set of stage-specific action features that are strongly correlated with each behavioral stage. Other methods can also be used in other embodiments, which are not limited here.
[0032] In specific implementation, organizing all stage-specific action features into action pattern features of beef cattle in different behavioral stages according to time sequence can be achieved in the following way: all stage-specific action features aligned by timestamps can be spliced into a time-series feature matrix in time sequence. The time-series feature matrix is then dynamically time-normalized and aligned to eliminate the time scale differences of different cycle behaviors. Then, the feature sequences are unified to the same length through interpolation methods. The final output is the action pattern feature sequence of beef cattle in different behavioral stages. This sequence completely represents the action evolution process of beef cattle in different behavioral stages. Other methods can also be used to determine this in other embodiments, which are not limited here.
[0033] It should be noted that the behavioral stage label in this application refers to the classification identifier used to discretize the continuous behavioral flow of beef cattle to identify its current dominant activity type; the stage-specific action features in this application refer to the set of parameters used to quantitatively describe the representative action details exhibited by beef cattle in different behavioral stages; and the action pattern features of different behavioral stages in this application refer to the characteristic sequence of the action evolution law and rhythm of individual beef cattle between different behavioral stages.
[0034] In some embodiments, generating the dynamic trajectory of beef cattle behavior in a beef cattle farming area based on the action pattern characteristics of the different behavioral stages can be achieved by the following steps: Based on the aforementioned action pattern features, an initial behavioral state sequence of beef cattle in the beef cattle breeding area is constructed; The initial behavior state sequence is subjected to time-series smoothing to generate an optimized state sequence; The dynamic trajectory of beef cattle behavior in the beef cattle breeding area is generated based on the optimized state sequence.
[0035] In specific implementation, the initial behavioral state sequence of beef cattle in the beef cattle breeding area can be constructed based on the action pattern features in the following way: the feature vector of each time point in the action pattern feature sequence can be used as the behavioral state vector of the beef cattle at the corresponding time; these behavioral state vectors are arranged in order according to their corresponding timestamps to form the initial behavioral state sequence of beef cattle in the beef cattle breeding area; wherein, each state vector in the sequence encapsulates the comprehensive state of the beef cattle behavior at that time represented by the action pattern features. Other methods can also be used in other embodiments, which are not limited here.
[0036] In specific implementation, the initial behavioral state sequence is subjected to temporal smoothing to generate an optimized state sequence. This can be achieved by inputting the initial behavioral state sequence into a Kalman filter. This filter is based on the system's state-space model and uses the state estimate from the previous moment to predict the state at the current moment. It then performs a weighted fusion of this predicted value with the observed value at the current moment, i.e., the state vector in the initial behavioral state sequence, thereby correcting abrupt changes in the state estimation caused by instantaneous noise or observation errors. After the filter recursively estimates and corrects the state at each moment in the sequence, it outputs an optimized state sequence that is smooth and continuous in the time dimension and free from instantaneous disturbances. Other methods can also be used to determine this in other embodiments, which are not limited here.
[0037] In specific implementation, the generation of the dynamic behavioral trajectory of beef cattle in the beef cattle breeding area based on the optimized state sequence can be achieved in the following way: the output of the dynamic behavioral trajectory of beef cattle in the beef cattle breeding area can be generated based on the optimized state sequence; the dynamic behavioral trajectory is a continuously optimized state vector sequence, where the sequence order represents the passage of time, and the evolution of each state vector in the sequence directly and continuously reflects the dynamic change process of the individual behavioral pattern of beef cattle in the beef cattle breeding area, providing a coherent time-series data foundation for fine-grained identification of health change trends. Other methods can also be used to determine this in other embodiments, which are not limited here.
[0038] It should be noted that the initial behavioral state sequence in this application refers to the initial set of beef cattle behavioral state vectors arranged in chronological order obtained by action pattern feature mapping; the optimized state sequence in this application refers to the state vector sequence that can truly reflect the inherent change law of beef cattle behavior after denoising and smoothing the initial behavioral state sequence; and the behavioral dynamic trajectory in this application refers to the trajectory that characterizes the temporal change law of beef cattle behavior patterns and growth and health status.
[0039] In step 103, the health change trend of beef cattle in different breeding periods is identified in a fine-grained segment based on the dynamic behavioral trajectory, and the local feature vector of the health status of beef cattle in each breeding period is obtained. The behavioral stability of health monitoring of beef cattle in different breeding periods is determined by all the local feature vectors.
[0040] In some embodiments, reference Figure 3 As shown, this figure is an exemplary flowchart of fine-grained segmented recognition in some embodiments of this application. Fine-grained segmented recognition of the health change trends of beef cattle at different breeding periods based on the dynamic behavioral trajectory, and obtaining local feature vectors of the health status of beef cattle at each breeding period, can be achieved through the following steps: The dynamic trajectory of the behavior is segmented into fine-grained segments based on the preset breeding time period to obtain the trajectory segments corresponding to each breeding time period; Statistical and frequency domain features characterizing the health trends of beef cattle are extracted from trajectory segments of each breeding period. The extracted statistical features and frequency domain features are fused into a local feature vector of the health status of beef cattle in the corresponding breeding period.
[0041] In specific implementation, the dynamic trajectory of behavior is finely segmented based on a preset breeding period to obtain the trajectory segments corresponding to each breeding period. This can be achieved in the following way: the dynamic trajectory of behavior can be divided into multiple continuous time segments according to a fixed time window, such as 24 hours as a breeding period, or according to the behavioral rhythm cycle identified in the dynamic trajectory of behavior, such as a complete feeding-resting cycle as a preset breeding period, so as to obtain the trajectory segments corresponding to each breeding period. Each trajectory segment can contain all state vectors arranged in chronological order within that breeding period, and the time span of each segment completely corresponds to the preset breeding period. Other methods can also be used in other embodiments, which are not limited here.
[0042] In specific implementation, the statistical and frequency domain features representing the health change trend of beef cattle can be extracted from the trajectory segments of each breeding period in the following way: For each trajectory segment of the breeding period, the mean, variance, and extreme values of its state vector in each dimension can be calculated as statistical features representing the health change trend of beef cattle; at the same time, a fast Fourier transform is performed on the trajectory segment to extract the dominant frequency component and corresponding energy in its amplitude spectrum as frequency domain features representing the health change trend of beef cattle; wherein, the statistical features reflect the central tendency and dispersion of the beef cattle behavior pattern, and the frequency domain features reveal the periodicity and rhythmic pattern of the beef cattle behavior; other methods can also be used in other embodiments, which are not limited here.
[0043] In specific implementation, the fusion of extracted statistical features and frequency domain features into a local feature vector of beef cattle health status for the corresponding breeding period can be achieved in the following way: the statistical features and frequency domain features extracted for each breeding period, namely mean, variance, extreme values, dominant frequency components, and energy, can be concatenated into a comprehensive feature vector. This vector is then subjected to max-min normalization, and the final output is a standardized local feature vector, thus obtaining the local feature vector of beef cattle health status for each breeding period. Each local feature vector uniquely represents the comprehensive state of beef cattle health and behavior patterns within the corresponding breeding period. Other methods can also be used to determine this in other embodiments, which are not limited here.
[0044] It should be noted that, in this application, the breeding period refers to a basic observation unit with a predetermined time length set for health trend analysis of beef cattle breeding areas; the trajectory segment in this application refers to a continuous state vector segment corresponding to the behavioral dynamic trajectory within a single breeding period, which is used to carry complete temporal information on the behavior and health status of beef cattle within a specified breeding period; and the local feature vector in this application refers to a standardized set of parameters that characterize the key features of the health and behavioral patterns of beef cattle within a single breeding period.
[0045] In some embodiments, determining the behavioral stability of beef cattle health monitoring at different breeding periods using all local feature vectors can be achieved through the following steps: All local feature vectors are organized into a temporal feature sequence in chronological order; Determine the similarity matrix between adjacent feature vectors in the time-series feature sequence; The similarity matrix was used to determine the behavioral stability of beef cattle health monitoring at different breeding periods.
[0046] In specific implementation, organizing all local feature vectors into a temporal feature sequence in chronological order can be achieved in the following way: the local feature vectors can be arranged sequentially according to the chronological order of the breeding periods corresponding to each local feature vector to form an ordered sequence, and this sequence can be used as the temporal feature sequence; each element in this sequence is a local feature vector, and its order truly reflects the temporal evolution of beef cattle health and behavior patterns in different breeding periods. Other methods can also be used to determine this in other embodiments, which are not limited here.
[0047] In specific implementation, the similarity matrix between adjacent feature vectors in the time-series feature sequence can be determined in the following way: the cosine similarity between the local feature vectors of every two adjacent breeding periods in the time-series feature sequence can be calculated sequentially; the similarity values of all adjacent breeding periods can be arranged into a vector in chronological order, and the vector can be organized into a one-dimensional similarity matrix; wherein each similarity value quantifies the continuity and degree of change of the health and behavior patterns of beef cattle between adjacent breeding periods; other methods can also be used to determine this in other embodiments, which are not limited here.
[0048] In specific implementation, the behavioral stability of beef cattle health monitoring at different breeding periods can be determined based on the similarity matrix in the following way: the arithmetic mean of all similarity values in the similarity matrix can be calculated, and this mean can be used as an indicator of the behavioral stability of beef cattle health monitoring at different breeding periods; the higher the value of this indicator, the more stable the behavioral pattern of beef cattle at different breeding periods and the smoother the change in health status, while the lower the value, the more likely there are fluctuations or abnormalities in the health status of beef cattle; other methods can also be used in other embodiments, which are not limited here.
[0049] It should be noted that the temporal feature sequence in this application refers to a set of local feature vectors that reflect the evolution of beef cattle health and behavior patterns over time; the similarity matrix in this application refers to a set of similarity values used to quantify the continuity of beef cattle health and behavior patterns between adjacent breeding periods; and the behavioral stability in this application refers to a quantitative indicator used to comprehensively evaluate the stability of the health status and the regularity of the behavior patterns of individual beef cattle in different breeding periods.
[0050] In step 104, based on the correlation changes between the behavioral stability and the environmental data of the beef cattle population, a visual risk index of the health status of beef cattle in the beef cattle breeding area is determined.
[0051] In some embodiments, determining a visualized risk index of beef cattle health status in a beef cattle farming area based on the correlation changes between behavioral stability and beef cattle herd environmental data can be achieved through the following steps: Obtain environmental data for beef cattle herds; Determine the correlation and change patterns between the behavioral stability and environmental data; Based on the aforementioned correlation change patterns, a visualized risk index is determined for the health status of beef cattle in beef cattle farming areas, including those with malnutrition, abnormal movement, suspected diseases, and stress responses.
[0052] In specific implementation, the environmental data of the beef cattle herd can be obtained in the following way: the environmental data of the beef cattle herd can be collected in real time by IoT sensor nodes deployed in the beef cattle breeding area. The parameters include, but are not limited to, pen temperature, humidity, ammonia concentration and stocking density. The temperature, humidity and ammonia concentration are directly read by the corresponding digital sensors. The stocking density is calculated by counting the number of individual beef cattle in the current breeding area obtained by image analysis and combining it with the known area. After data cleaning and format standardization, all parameters are stored together with the environmental data collection timestamp. Other methods can also be used to obtain the data in other embodiments, which are not limited here.
[0053] In specific implementation, determining the correlation change pattern between behavioral stability and environmental data can be achieved in the following way: First, the behavioral stability and environmental data of the current and historical breeding periods can be aligned according to a time window; then, a sliding window can be used to calculate the Pearson correlation coefficient between the trend of environmental data change and the trend of behavioral stability change, where the trend of environmental data change is such as the slope of temperature change, and the trend of behavioral stability change is such as the difference sequence of behavioral stability; finally, by analyzing the sign, magnitude and evolution of the correlation coefficient over time, the dynamic correlation change pattern such as positive correlation, negative correlation or no significant correlation between the two can be determined, thereby quantifying the degree of influence of environmental load on behavioral stability. Other methods can also be used to determine this in other embodiments, which are not limited here.
[0054] In specific implementation, the visualized risk index for determining the health status of beef cattle in a beef cattle breeding area, including those with malnutrition, abnormal movement, suspected disease, and stress response, based on the aforementioned correlation change pattern, can be achieved in the following way: The correlation change pattern, including correlation coefficients and trends, is input into a risk decision tree predefined based on historical data and expert knowledge for pattern matching. This decision tree includes rules such as: if behavioral stability remains consistently low and has no significant correlation with environmental data, it is determined to be suspected disease; if decreased behavioral stability shows a strong negative correlation with increased ammonia concentration, it is determined to be a stress response; if... If the stability decreases slowly and the stocking density is normal, it is judged as malnutrition; if the gait-related features in the behavioral stability fluctuate drastically, it is judged as movement abnormality, where the gait-related features include the amplitude of the gait movement vector; finally, based on the judged risk type and combined with the current value of behavioral stability, it is mapped to a specific value within a standardized range, such as a scale of 0-100, through linear interpolation, thus serving as a visualized risk index corresponding to the health status of beef cattle in beef cattle breeding areas, such as malnutrition, movement abnormality, suspected disease, and stress response; other methods can also be used to determine this in other embodiments, which are not limited here.
[0055] It should be noted that the environmental data in this application refers to a set of external factor indicators used to quantify the overall environmental pressure and carrying capacity of beef cattle farming areas. These indicators serve as key input variables for assessing the impact of the external environment on the health and behavioral patterns of beef cattle herds. The correlation change pattern in this application refers to a quantitative representation used to reveal the dynamic interaction between the stability of individual beef cattle behavior and the environmental pressure of the herd. This pattern serves as a core criterion for distinguishing different health risk sources and establishing a causal relationship between environmental factors and behavioral performance. The visualized risk index in this application refers to a standardized numerical indicator that comprehensively reflects the types and severity of individual health risks in beef cattle within the beef cattle farming area.
[0056] In step 105, monitoring results for livestock management in beef cattle farming areas are generated based on the visualized risk index.
[0057] In specific implementation, the monitoring results for cattle breeding management based on the visualized risk index can be generated in the following way: First, the visualized risk index can be divided into different risk levels according to a preset threshold range, for example: 0-30 is low risk, 31-70 is medium risk, and 71-100 is high risk. Corresponding management suggestions are matched for each level, such as continuous observation for low risk, focused inspection for medium risk, and immediate intervention for high risk. Then, a structured monitoring report is automatically generated, organized in JSON format, including risk type, risk level, specific value, trend analysis, and corresponding management measures suggestions. Finally, the monitoring results are pushed to the breeding management terminal via a RESTful API interface. After receiving the report, the terminal displays it visually through a web interface in the form of a color-coded risk heat map and a graded alarm list. Simultaneously, SMS and voice alarms are automatically triggered for high-risk situations, ultimately forming the monitoring results output for cattle breeding management. Other methods can also be used in other embodiments, which are not limited here.
[0058] In another aspect, in some embodiments, this application provides a beef cattle farming monitoring system, which includes a visual monitoring unit, with reference to... Figure 4 The figure is a schematic diagram of the structure of a visual monitoring unit according to some embodiments of this application. The visual monitoring unit 400 includes: a data acquisition module 401, a processing module 402, and an execution module 403, which are described below: The acquisition module 401 in this application is mainly used to acquire real-time image sequences of individual beef cattle in the beef cattle breeding area, and to construct a multi-dimensional visual representation of the beef cattle target based on the real-time image sequences. Processing module 402, in this application, is mainly used to extract the action pattern features of beef cattle at different behavioral stages from the multidimensional visual representation, and generate the dynamic trajectory of beef cattle behavior in the beef cattle breeding area based on the action pattern features of the different behavioral stages. The processing module 402 described in this application is also used to perform fine-grained segmentation identification of the health change trend of beef cattle in different breeding periods based on the behavioral dynamic trajectory, obtain local feature vectors of the health status of beef cattle in each breeding period, and determine the behavioral stability of beef cattle health monitoring in different breeding periods through all local feature vectors. The processing module 402 described in this application is also used to determine a visual risk index of the health status of beef cattle in the beef cattle breeding area based on the correlation changes between the behavioral stability and the environmental data of the beef cattle population; The execution module 403 in this application is mainly used to generate monitoring results for the management of beef cattle breeding areas based on the visualized risk index.
[0059] The foregoing has detailed examples of the beef cattle farming monitoring system and method provided in the embodiments of this application. It is understood that the corresponding apparatus, in order to achieve the above functions, includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0060] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described visual monitoring method for beef cattle farming status.
[0061] In some embodiments, reference Figure 5 The dashed lines in the figure indicate that the unit or module is optional. This figure is a schematic diagram of the computer device used to implement the visual monitoring method for beef cattle farming status according to this application. The visual monitoring method for beef cattle farming status in the above embodiments can be achieved through… Figure 5 The computer device 500 shown is used to implement this, and the computer device 500 includes at least one processor 501, a memory 502 and at least one communication unit 505. The computer device 500 may be a terminal device, a server or a chip.
[0062] The processor 501 can be a general-purpose processor or a special-purpose processor. For example, the processor 501 can be a central processing unit (CPU). The CPU can be used to control the computer device 500, execute software programs, and process data from the software programs. The computer device 500 may also include a communication unit 505 for inputting (receiving) and outputting (transmitting) signals.
[0063] For example, computer device 500 may be a chip, communication unit 505 may be the input and / or output circuit of the chip, or communication unit 505 may be the communication interface of the chip, and the chip may be a component of terminal device, network device or other device.
[0064] For example, computer device 500 may be a terminal device or a server, and communication unit 505 may be a transceiver of the terminal device or the server, or communication unit 505 may be a transceiver circuit of the terminal device or the server.
[0065] The computer device 500 may include one or more memories 502 storing a program 504. The program 504 can be executed by a processor 501 to generate instructions 503, causing the processor 501 to perform the methods described in the above method embodiments according to the instructions 503. Optionally, the memory 502 may also store data (such as a target audit model). Optionally, the processor 501 may also read data stored in the memory 502, which may be stored at the same storage address as the program 504, or the data may be stored at a different storage address than the program 504.
[0066] The processor 501 and memory 502 can be configured separately or integrated together, for example, integrated on the system-on-chip (SOC) of the terminal device.
[0067] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 501. The processor 501 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gates, transistor logic devices, or discrete hardware components.
[0068] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0069] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described visual monitoring method for the state of beef cattle farming.
[0070] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.
[0071] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.
Claims
1. A visual monitoring method for the status of beef cattle farming, applied to a beef cattle farming monitoring system, characterized in that, The method includes the following steps: Real-time image sequences of individual beef cattle in the beef cattle breeding area are collected, and a multi-dimensional visual representation of the beef cattle target is constructed based on the real-time image sequences; The action pattern features of beef cattle at different behavioral stages are extracted from the multidimensional visual representation, and the dynamic trajectory of beef cattle behavior in the beef cattle breeding area is generated based on the action pattern features at different behavioral stages. Based on the aforementioned behavioral dynamic trajectory, the health change trend of beef cattle in different breeding periods is identified in a fine-grained segmentation, and the local feature vector of the health status of beef cattle in each breeding period is obtained. The behavioral stability of health monitoring of beef cattle in different breeding periods is determined by all the local feature vectors. Based on the correlation changes between the behavioral stability and the environmental data of the beef cattle population, a visual risk index of the health status of beef cattle in the beef cattle breeding area is determined; Based on the visualized risk index, monitoring results are generated for livestock management in beef cattle breeding areas.
2. The method as described in claim 1, characterized in that, Constructing a multidimensional visual representation of beef cattle targets based on the real-time image sequence specifically includes: Individual cattle detection and segmentation are performed on the real-time image sequence; Based on the segmentation results, the posture key points, gait motion vectors and body surface contour features of individual beef cattle were extracted. Determine the relative position matrix among individual beef cattle within the herd; The posture key points, gait motion vectors, body surface contour features, and relative position matrix are fused to generate a multidimensional visual representation of the beef cattle target.
3. The method as described in claim 1, characterized in that, Extracting the movement pattern features of beef cattle at different behavioral stages from the multidimensional visual representation specifically includes: Behavioral stage identification is performed on the multidimensional visual representation to obtain behavioral stage labels for beef cattle, including feeding, movement, and rest. Extract stage-specific action features based on labels for each behavioral stage; All stage-specific action characteristics were organized chronologically into action pattern characteristics of beef cattle at different behavioral stages.
4. The method as described in claim 1, characterized in that, Generating the dynamic trajectory of beef cattle behavior in a beef cattle farming area based on the action pattern characteristics of different behavioral stages specifically includes: Based on the aforementioned action pattern features, an initial behavioral state sequence of beef cattle in the beef cattle breeding area is constructed; The initial behavior state sequence is subjected to time-series smoothing to generate an optimized state sequence; The dynamic trajectory of beef cattle behavior in the beef cattle breeding area is generated based on the optimized state sequence.
5. The method as described in claim 1, characterized in that, Based on the aforementioned behavioral dynamic trajectory, fine-grained segmentation identification of the health change trend of beef cattle at different breeding periods is performed to obtain the local feature vector of the health status of beef cattle at each breeding period, specifically including: The dynamic trajectory of the behavior is segmented into fine-grained segments based on the preset breeding time period to obtain the trajectory segments corresponding to each breeding time period; Statistical and frequency domain features characterizing the health trends of beef cattle are extracted from trajectory segments of each breeding period. The extracted statistical features and frequency domain features are fused into a local feature vector of the health status of beef cattle in the corresponding breeding period.
6. The method as described in claim 1, characterized in that, Determining the behavioral stability of beef cattle health monitoring at different breeding stages by using all local feature vectors specifically includes: All local feature vectors are organized into a temporal feature sequence in chronological order; Determine the similarity matrix between adjacent feature vectors in the time-series feature sequence; The similarity matrix was used to determine the behavioral stability of beef cattle health monitoring at different breeding periods.
7. The method as described in claim 1, characterized in that, Based on the correlation changes between the behavioral stability and the environmental data of the beef cattle population, the visualized risk index of the health status of beef cattle in the beef cattle breeding area is determined, specifically including: Obtain environmental data for beef cattle herds; Determine the correlation and change patterns between the behavioral stability and environmental data; Based on the aforementioned correlation change patterns, a visualized risk index is determined for the health status of beef cattle in beef cattle farming areas, including those with malnutrition, abnormal movement, suspected diseases, and stress responses.
8. A beef cattle farming monitoring system, comprising a visual monitoring unit, characterized in that, The visual monitoring unit includes: The acquisition module is used to acquire real-time image sequences of individual beef cattle in the beef cattle breeding area, and to construct a multi-dimensional visual representation of the beef cattle target based on the real-time image sequences. The processing module is used to extract the action pattern features of beef cattle at different behavioral stages from the multidimensional visual representation, and generate the dynamic trajectory of beef cattle behavior in the beef cattle breeding area based on the action pattern features at different behavioral stages. The processing module is also used to perform fine-grained segmentation identification of the health change trend of beef cattle in different breeding periods based on the behavioral dynamic trajectory, obtain local feature vectors of the health status of beef cattle in each breeding period, and determine the behavioral stability of health monitoring of beef cattle in different breeding periods through all local feature vectors. The processing module is also used to determine a visual risk index of the health status of beef cattle in the beef cattle breeding area based on the correlation changes between the behavioral stability and the environmental data of the beef cattle population. The execution module is used to generate monitoring results for livestock management in beef cattle breeding areas based on the visualized risk index.
9. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store computer programs, and the processor is used to call and run the computer programs from the memory, so that the computer device performs the visual monitoring method for beef cattle breeding status as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions or code that, when executed on a computer, cause the computer to implement the visual monitoring method for the state of beef cattle farming as described in any one of claims 1 to 7.