Chicken flock health monitoring method, system and storage medium

Abstract

This invention discloses a method, system, and storage medium for monitoring the health of chicken flocks, including: acquiring video images of the chicken flock's activity area; generating distribution data based on the video images; extracting distribution feature parameters from the distribution data to describe the spatial distribution characteristics of the chicken flock's movement intensity; comparing these parameters with baseline parameters; and issuing a health warning signal when abnormal deviations occur. This invention calculates the spatial distribution data of the chicken flock's movement simply by comparing images, requiring minimal computation. Furthermore, this invention monitors the entire chicken flock, completely independent of target detection and multi-target tracking technologies, avoiding tracking failures in scenarios with high stocking density and chickens obscuring each other. This invention acquires video images via a camera, providing 24-hour coverage, and calculates real-time monitoring data for the chicken flock. Abnormal deviations in the monitoring data trigger real-time health warning signals, ensuring high timeliness and avoiding subjective judgment differences inherent in manual inspections.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology in livestock farming, and in particular to a method, system and storage medium for monitoring the health of chicken flocks. Background Technology

[0002] In modern poultry farming, flock health monitoring is a crucial link in ensuring production efficiency and animal welfare. Large-scale poultry farming relies primarily on manual inspections due to cost considerations. However, manual inspections suffer from drawbacks such as high latency, inability to provide 24-hour coverage, and strong subjectivity. Often, diseases are only detected when sick chickens show obvious symptoms, by which time the disease may have already spread throughout the flock. Automated monitoring mainly relies on individual sensor monitoring or deep learning-based video analytics. Individual sensor monitoring requires attaching sensors to each chicken, which is costly, cumbersome, and prone to stress. Deep learning-based video analytics utilizes object detection and multi-object tracking algorithms to identify and track the movement trajectory of each chicken and extract individual behavioral indicators. These solutions are algorithmically complex, require high-performance GPU servers, are costly, and have a high failure rate in scenarios with high stocking density and chickens occluding each other. Summary of the Invention

[0003] The main objective of this invention is to provide a method, system, and storage medium for monitoring the health of chicken flocks. This invention aims to address the problems of existing technologies, such as the high lag in manual inspections, the inability to provide 24-hour coverage, the high degree of subjectivity, the high cost and cumbersome process of individual sensor solutions requiring each chicken to wear a sensor, and the high cost of deep learning-based video analysis monitoring solutions requiring high-performance GPU servers and high cost, as well as the high failure rate in scenarios with high stocking density and chickens occluding each other.

[0004] In a first aspect, to achieve the above objectives, the present invention provides a method for monitoring the health of a chicken flock, comprising:

[0005] Collect video images of the chickens' activity area;

[0006] Based on the video images, distribution data reflecting the spatial distribution of the chicken flock's movement intensity is generated;

[0007] Based on the distribution data, distribution feature parameters are extracted to describe the spatial distribution characteristics of the chicken flock's movement intensity;

[0008] Compare the current distribution characteristic parameters with the corresponding baseline parameters established under historical health conditions;

[0009] When the comparison results indicate that the current distribution characteristic parameter deviates abnormally from the baseline parameter, a health warning signal is issued.

[0010] Optionally, the distribution characteristic parameter includes kurtosis; the abnormal deviation includes an increase in kurtosis relative to the kurtosis baseline in the baseline parameter exceeding a predetermined threshold.

[0011] Optionally, the distribution characteristic parameters further include skewness; the health warning signal includes a hybrid anomaly signal, which is issued when the skewness does not exceed the preset range of the skewness baseline in the baseline parameters but the kurtosis deviates abnormally.

[0012] Optionally, the baseline parameters are established separately according to multiple pre-divided time periods throughout the day.

[0013] Optionally, the baseline parameters are dynamically updated using a sliding time window or an exponentially weighted moving average.

[0014] Optionally, the step of generating distribution data reflecting the spatial distribution of the chicken flock's movement intensity based on the video image includes:

[0015] The motion information of the chicken flock is generated based on the video images; wherein, the motion information is obtained by the inter-frame difference method: the pixel grayscale difference between two adjacent frames or two frames with a predetermined interval in the video image sequence is calculated as the motion information;

[0016] The distribution data is generated based on the motion information.

[0017] Optionally, the step of generating distribution data reflecting the spatial distribution of the chicken flock's movement intensity based on the video image includes:

[0018] The video image is divided into multiple spatial grids, and the statistical value of motion intensity within each grid is calculated. The set of these statistical values ​​is used as the distribution data.

[0019] Optionally, it further includes: identifying low-motion regions where the motion intensity is below a predetermined threshold, tracking the displacement of the center of mass of the region over a continuous time period, and issuing the warning signal when the displacement is less than the predetermined threshold and the duration exceeds a predetermined value.

[0020] Secondly, the present invention also provides a chicken flock health monitoring system, comprising:

[0021] The video image acquisition module is used to acquire video images of the area where the chickens are active.

[0022] A processor configured to perform the flock health monitoring method according to any one of claims 1 to 8;

[0023] The alarm module is used to issue an alarm when the processor sends a health warning signal.

[0024] Thirdly, the present invention also provides a computer-readable storage medium storing a flock health monitoring program, which, when executed by a processor, implements the flock health monitoring method as described in any one of claims 1 to 8.

[0025] This invention calculates the spatial distribution of chicken movement by comparing images, requiring minimal computation, no high-performance GPU servers, no individual data, and no sensors on the chickens, thus reducing costs and avoiding the stress caused by sensor-equipped chickens. Furthermore, this invention monitors the entire flock, completely independent of target detection and multi-target tracking technologies, avoiding tracking failures in high-density environments where chickens obstruct each other. By capturing video images through cameras, this invention provides 24-hour coverage and calculates real-time monitoring data for the flock. A health warning signal is only issued in real time when abnormal deviations in the monitoring data occur, ensuring high timeliness and eliminating the problem of human subjectivity. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0027] Figure 1 This is a flowchart illustrating a method for monitoring chicken health according to an embodiment of the present invention.

[0028] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0029] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0030] In this invention, the use of terms such as "first," "second," etc., is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0031] This invention provides a first embodiment of a method for monitoring the health of chicken flocks. In this embodiment, as shown... Figure 1 As shown, the chicken flock health monitoring method of the present invention includes the following steps:

[0032] Step S100: Collect video images of the chickens' activity area;

[0033] Cameras are installed 2.5m-4m above the area where the chickens need to be monitored. One camera can typically monitor 50 to 300 chickens. The placement of cameras depends on the scale of the farm and the layout of the chicken coop; this invention does not impose specific limitations. The acquisition frequency can be set as needed. Understandably, a higher acquisition frequency results in more accurate data generated from the video images; a lower acquisition frequency is more energy-efficient. In practical use, based on the chickens' habits, a camera is typically selected to acquire video images once per minute.

[0034] Step S200: Generate distribution data reflecting the spatial distribution of the chicken flock's movement intensity based on the video images;

[0035] Understandably, by comparing images, the magnitude of change at each pixel location can be calculated. Large changes within a region indicate that chickens are moving rapidly at that location, while small or unchanged changes indicate that chickens are stationary or only slightly moving. By representing the magnitude of change at each pixel and binding it to its location, the spatial distribution data of the chicken movement intensity can be obtained. To reduce data volume and computational complexity, the image can be divided into multiple sub-regions (e.g., a grid), and the average movement intensity within each sub-region can be used as the distribution data value for that location.

[0036] Step S300: Extract distribution feature parameters from the distribution data to describe the spatial distribution characteristics of the chicken flock's movement intensity.

[0037] It should be noted that the distribution characteristic parameters in this invention refer to statistical quantities that can quantify the spatial distribution pattern of motion intensity, including but not limited to: skewness, kurtosis, spatial entropy, variance, coefficient of variation, Gini coefficient, etc. Those skilled in the art will understand that any parameter that can reflect the distribution shape, dispersion, uniformity, or clustering of motion intensity on a spatial grid can be used as the distribution characteristic parameter described in this invention. Those skilled in the art can select appropriate parameters according to actual needs.

[0038] Step S400: Compare the current distribution characteristic parameters with the corresponding baseline parameters established under historical health conditions;

[0039] It should be noted that the baseline parameters established under historical health conditions can be formed from data of other batches of healthy chickens of the same breed and age, or from data collected over a period of time from newly arrived chickens of the same batch (whose health status has been confirmed by other testing methods). Baseline parameters include the normal range of values ​​for this distributional characteristic parameter under healthy conditions (such as mean, standard deviation, etc.). When the system is first deployed or a new flock is introduced, an initial learning period (e.g., 72 consecutive hours) is required. During this learning period, the system only collects data and establishes baseline parameters, without issuing health warning signals. After the learning period ends, the system automatically activates the warning function. If significant data fluctuations occur during the learning period (e.g., persistently abnormally high or low exercise intensity), the system can prompt the farmer to check the flock's status. After confirming health, the learning period can be reset, or the collected data can be used as a valid baseline.

[0040] Step S500: When the comparison results indicate that the current distribution characteristic parameter deviates abnormally from the baseline parameter, a health warning signal is issued.

[0041] Understandably, abnormal deviations can refer to either abnormally low or abnormally high activity levels. Numerous animal behavior and veterinary studies have shown that abnormally low activity levels are an important early warning sign in poultry farming. When a chicken suddenly changes from normal activity to prolonged periods of stillness, it is highly likely that there is a problem with its physiological functions. For example, chickens suffering from respiratory diseases (infectious bronchitis, avian influenza, mycoplasma) may exhibit symptoms such as standing still, retracting their necks, closing their eyes, and being unwilling to move; chickens suffering from digestive tract diseases (coccidiosis, enteritis) may exhibit symptoms such as lethargy, drooping wings, and crouching away from the flock; chickens suffering from leg diseases (staphylococcal arthritis, leg weakness) may exhibit symptoms such as lameness and lying down; and chickens with systemic infections (E. coli, Salmonella) may exhibit symptoms such as loss of appetite, ruffled feathers, and lethargy with closed eyes. All these diseases will lead to a significant decrease in the chicken's activity level. Other situations, such as stress, can cause chickens to become restless, pacing, flapping their wings, and squawking; skin parasites can cause intense itching, leading to constant feather pecking and scratching; certain neurological symptoms can stimulate the chicken's central nervous system, resulting in aimless movements and abnormal activity patterns; and unsuitable environments can cause restlessness, with overall activity levels increasing and becoming irregular. These conditions or diseases can all significantly increase the intensity of chicken activity. Therefore, when comparative distribution characteristic parameters deviate abnormally, a health warning signal should be issued to prompt human intervention and achieve early disease warning. Specifically, the signal can be set to only be issued when the same abnormal deviation occurs continuously for a period of time (e.g., more than twenty minutes), reducing false alarms caused by instantaneous changes in the flock.

[0042] This invention calculates the spatial distribution of chicken movement by comparing images, requiring minimal computation, no high-performance GPU servers, low computing power, energy conservation, and emission reduction. It also eliminates the need for individual data and sensors on chickens, reducing costs and avoiding the stress associated with sensor-equipped chickens. Furthermore, this invention monitors the entire flock, completely independent of target detection and multi-target tracking technologies, avoiding tracking failures in high-density environments where chickens obstruct each other. By capturing video images through cameras, it provides 24-hour coverage and calculates real-time monitoring data for the flock. Abnormal deviations in the monitoring data trigger immediate health warnings, preventing delays until obvious symptoms appear, ensuring high timeliness and avoiding subjective judgment errors inherent in manual inspections.

[0043] In one embodiment, the distribution characteristic parameter includes kurtosis; abnormal deviation includes an increase in kurtosis relative to the kurtosis baseline in the baseline parameter exceeding a predetermined threshold.

[0044] For a sample set X = {x1, x2, ..., x...} consisting of the motion intensity values ​​of each region... N The kurtosis value K is calculated using the following formula (where N is the number of regions):

[0045]

[0046] in, The mean of the sample set data. denoted as the standard deviation of the sample set data.

[0047] In this embodiment, the kurtosis baseline is the mean of the kurtosis data for each time period within the recorded healthy period. (Or the median can be used). When the current kurtosis K is relative to the kurtosis baseline When the activity level of chickens exceeds a predetermined threshold, it indicates an abnormal increase, decrease, or both. A health warning signal is then issued to prompt human intervention. Specifically, the predetermined threshold can be set to... The ratio is 1.2 times, 1.5 times, or 2 times, with 1.5 times being preferred in this embodiment.

[0048] Kurtosis is highly sensitive to outliers in the distribution of activity intensity. Kurtosis increases significantly when sick chickens (extremely low activity levels) or stressed / excited chickens (extremely high activity levels) are present in a flock; especially when both types of anomalies coexist, kurtosis can still be reliably detected, while the mean, variance, and even skewness may show no significant changes. Therefore, using kurtosis as a distribution characteristic parameter can effectively identify mixed anomalies that traditional indicators easily miss, improving the comprehensiveness and accuracy of early warning systems.

[0049] Furthermore, the distribution characteristic parameters also include skewness; the health warning signal includes a mixed-type abnormal signal, which is issued when the skewness does not exceed the preset range of the skewness baseline in the baseline parameters but the kurtosis deviates abnormally.

[0050] For a sample set X = {x1, x2, ..., x...} consisting of the motion intensity values ​​of each region... N (N is the number of regions) The skewness value S is calculated using the following formula:

[0051]

[0052] in, The mean of the sample set data. denoted as the standard deviation of the sample set data.

[0053] Similarly, the skewness baseline is the mean of the skewness data for each time period within the recorded healthy period. (Alternatively, the median can be used), while simultaneously calculating and recording its dispersion. This allows us to obtain the normal range. Specifically, the skewness baseline can be set to the mean of the skewness data. Dispersion in healthy periods at positive and negative multiples The range below is preferably three times. ,Right now[ , Skewness alone can detect two situations: abnormally high activity levels or abnormally low activity levels in chickens. When the current kurtosis K is relative to the kurtosis baseline... When the increase exceeds a predetermined threshold and the current skewness S is within the normal range, it indicates that both abnormally high and abnormally low activity levels exist simultaneously, thus issuing mixed-type abnormal information. This enables the detection of situations where the total activity level remains unchanged, but individuals with both abnormally high and abnormally low activity levels exist simultaneously, which is difficult to detect under other observation modes of the entire flock, thereby improving the practicality of the invention.

[0054] Furthermore, it's understandable that combining kurtosis and skewness to monitor abnormal chicken activity yields higher accuracy. When the kurtosis (K) value is abnormally high and the skewness (S) value is negative, staff can clearly identify an abnormally low activity level and should search for inactive individuals in the flock. Similarly, when the kurtosis (K) value is abnormally high and the skewness (S) value is positively skewed, staff can clearly identify an abnormally high activity level and should search for hyperactive individuals in the flock. Conversely, when the kurtosis (K) value is abnormally high and the skewness (S) value is normal, both inactive and hyperactive individuals should be searched, greatly facilitating timely investigation by staff. Additionally, it's important to clarify that health alerts are issued based on an abnormal kurtosis (K) value; no health alert is issued if the kurtosis (K) value is normal but the skewness (S) value is abnormal. The skewness (S) value primarily serves to identify situations where both inactive and hyperactive individuals are present, and to clarify whether the abnormality specifically involves inactive or hyperactive individuals.

[0055] In one embodiment, baseline parameters are established according to multiple pre-divided time periods throughout the day. Understandably, based on the natural activity rhythm of the flock, the day is divided into the following time periods (which can be adjusted according to the actual breed and season): Typically, from 5:00 AM to 7:00 AM, the chickens are just waking up, and their activity level gradually increases; the baseline parameters can be adjusted accordingly to gradually increase these periods. From 7:00 AM to 9:00 AM is the peak feeding time in the morning, and the flock is highly active; therefore, any abnormal decrease in activity level should be noted, and the threshold setting of the baseline parameters should be more sensitive. Specifically, the negative range of the skewness value can be reduced, and the positive range increased. Typically, from 12:00 PM to 2:00 PM and from 9:00 PM to 5:00 AM the next day are the afternoon rest and nighttime sleeping times; the flock will lie down and sleep, and their activity level is very low. Any abnormal increase in activity level should be noted, and the negative range of the skewness value can be increased, and the positive range decreased. From 6:00 PM to 9:00 PM, the chickens gradually fall asleep, and the baseline parameters can be adjusted accordingly to gradually decrease these periods. In another embodiment, baseline data parameters and strategies for issuing health warnings can be adjusted specifically based on the extracted distribution feature parameters and the natural activity rhythm of the flock. Establishing baselines for different time periods can accurately distinguish between normal physiological behavior and pathological abnormalities, significantly reducing the false alarm rate.

[0056] Preferably, the system can store distribution characteristic parameters (skewness, kurtosis, mean, etc.) for each time period for an extended period, forming a health trend curve. Farmers can review historical data to observe changes in the flock's health status; for example, a slow, gradual increase in kurtosis may indicate the accumulation of chronic diseases, allowing for early intervention. This function helps upgrade from "early warning" to "health management."

[0057] Furthermore, the baseline parameters are dynamically updated using a sliding time window or an exponentially weighted moving average.

[0058] Exponential weighted average method:

[0059] The mean skewness baseline for each time period t Each time a new sample is collected, the current skewness S is calculated. current (Assuming the current data is not identified as abnormal) then update using the following formula:

[0060]

[0061] in, The smoothing factor has a value of 0.9 to 0.99, preferably 0.95.

[0062] Standard deviation The update uses a recursive formula:

[0063]

[0064] Sliding time window method:

[0065] Retain data for the most recent period (preferably 7 days), store data by time period each day, remove the oldest day at the end of each day, and recalculate the mean and standard deviation using data from the same time period.

[0066] Understandably, chicken activity patterns change slowly over time (e.g., chickens become more active as they grow, seasonal temperature changes occur, and the flock migrates). Using a fixed baseline for extended periods can lead to increased false alarms. Dynamic updates allow the baseline to automatically adapt to these normal changes, eliminating the need for frequent manual resets and improving the system's long-term stability and reliability.

[0067] In one embodiment, the step of generating distribution data reflecting the spatial distribution of the movement intensity of a flock of chickens based on video images includes: generating motion information of the flock of chickens based on video images; wherein the motion information is obtained by the inter-frame difference method: calculating the pixel grayscale difference between two adjacent frames or two frames with a predetermined interval in the video image sequence as motion information; and generating distribution data based on the motion information.

[0068] The camera can be set to capture grayscale images at a rate of 1 frame per second. For times t and t- ( The calculation of inter-frame differences can range from 1 second to 5 minutes (preferably 1 minute in this embodiment):

[0069]

[0070] D(x,y) represents the motion intensity of pixel (x,y), ranging from 0 to 255. The image is divided into a specific number of regions as needed. The motion intensity of each region is the average of the differences between all pixels within that region, resulting in the region motion intensity matrix X = {X...} i,jThis matrix represents the distributed data, and its calculation is simple, requiring only subtraction and absolute value operations. Compared to other calculation methods, the gradation of light has almost no impact on the inter-frame difference method, and it does not require background modeling. It has good real-time performance, can run on ordinary CPUs, and is low in cost and power consumption, making it more energy-efficient and emission-reducing.

[0071] It should be noted that in this invention, motion intensity is obtained through inter-frame differencing, which calculates the pixel grayscale difference between the current frame and the previous frame (or two frames separated by a predetermined number of frames). When sudden changes in light occur (such as manual switching on / off of lights or lightning), inter-frame differencing can produce large-area grayscale differences, causing a sudden increase in the total global motion intensity, potentially exceeding the healthy baseline range. However, the anomaly detection logic of this invention requires that the abnormal deviation of the current distribution feature parameters from the baseline parameters must persist for more than a predetermined time (e.g., 20 minutes) before a health warning signal is issued. Motion intensity anomalies caused by sudden changes in light, such as switching on / off lights or lightning, last only a very short time (usually no more than a few seconds), far below the 20-minute duration threshold. Therefore, such transient interference will not trigger the final warning signal, and the system does not need to design additional detection or filtering mechanisms for sudden changes in light. Simultaneously, after a sudden change in light, the reference frame for inter-frame differencing is automatically updated (the next frame is compared with the frame after the change), and the motion intensity data quickly returns to normal levels. Without additional processing, switching lights on and off or lightning strikes typically do not cause the system to generate abnormal deviation records lasting more than 20 minutes, thus not affecting the accuracy of health monitoring. The duration determination conditions of this invention naturally filter out transient external interference (including but not limited to sudden changes in light, occasional movement of chickens, and slight vibrations of the camera), ensuring the reliability of the invention.

[0072] In one embodiment, the step of generating distribution data reflecting the spatial distribution of chicken movement intensity based on video images includes: dividing the video image frame into multiple spatial grids, calculating the statistical value of movement intensity within each grid, and using the set of statistical values ​​as distribution data. After obtaining the movement intensity map, the image frame (or the cropped effective area) is spatially divided into R×C rectangular grids. For each grid, the statistical value of movement intensity of all pixels within that grid is calculated. This statistical value can be: arithmetic mean, median, or maximum value. Among these, the arithmetic mean has a relatively balanced function, the median has the advantage of noise resistance, and the maximum value is most sensitive to local intense movements. Therefore, this embodiment preferably uses the arithmetic mean as the statistical value. The statistical values ​​of all grids are arranged in spatial order to obtain the distribution data. Directly using pixel-level movement intensity will result in excessively high data dimensionality, and most of the calculations are invalid. Calculating statistics such as skewness and kurtosis is computationally intensive and easily affected by noise. Using a gridded approach compresses the data volume while retaining the necessary spatial distribution information.

[0073] Preferably, to reduce abrupt changes in movement intensity caused by chickens entering and exiting the edge of the field of vision, the original image can be cropped before analysis. For example, the central 90% of the image can be retained for analysis, while the outermost 5% of edge pixels can be ignored. Alternatively, after dividing the image into grids, the outermost grids can be assigned lower weights or excluded from the calculation of distribution feature parameters.

[0074] In one embodiment, the chicken flock health monitoring method further includes: identifying low-move regions with movement intensity below a predetermined threshold, tracking the displacement of the centroid of the region over a continuous time period, and issuing an early warning signal when the displacement is less than the predetermined threshold and the duration exceeds a predetermined value. Taking a gridded approach as an example, after obtaining the movement intensity distribution data (a gridded matrix), a global movement intensity threshold is set, for example, T = 0.2 × μ' (μ' is the average movement intensity of all current regions). For each grid, if its movement intensity is less than T, it is marked. Neighborhood connectivity analysis is performed, ignoring isolated noise points with an area less than two grids; each connected region is considered a low-activity cluster. For each connected region, its centroid coordinates are calculated:

[0075]

[0076] Among them, c i r is the index of the grid column in this connected region. i Let n be the index of the grid rows in the connected region, and n be the number of grids contained in the connected region. In a continuous time series (e.g., thirty minutes), match the same connected region by minimizing the centroid displacement. Record the total displacement of the region over the continuous time period:

[0077]

[0078] Where L represents the actual ground length corresponding to each grid cell. For example, calculating the displacement over thirty minutes... If the distance is less than 0.5m and the current time period is not within the normal allowable period for prolonged stillness, the area is determined to be stationary, a health warning signal is issued, and the grid coordinates of the area can be output for the breeder to view. Understandably, even when resting, healthy chickens will occasionally adjust their posture, move their position, or move slowly with the group, thus the centroid of low-movement areas will show significant displacement. Sick chickens (especially those with leg diseases or systemic infections) often remain motionless in the same position for extended periods. This embodiment, through centroid displacement tracking, can effectively distinguish between resting and stationary movement, further reducing the false alarm rate. It can also provide the location coordinates of suspected sick chickens, facilitating staff to quickly locate them. Specifically, this embodiment uses a gridded method to calculate the centroid, with sufficient accuracy to distinguish between stationary movement and resting. For scenarios requiring higher accuracy, the grid density can be increased or pixel-level connected components can be used directly, but this will increase the computational burden; those skilled in the art can choose according to actual needs.

[0079] Traditional flock health monitoring methods can only provide a qualitative judgment that "there is an abnormality in the flock," without providing the specific spatial location of the abnormal individual. When the system alarms, staff need to search through hundreds or even thousands of chickens one by one to find the abnormal individual. This process is time-consuming and labor-intensive, severely impacting the practical application value of the early warning system. To address this technical problem, this invention uses a centroid movement algorithm to calculate the centroid coordinates of connected regions based on the identification of low-motion areas, thereby outputting the approximate location of the abnormal individual. Although this method is limited by grid accuracy and cannot precisely locate one or two closely spaced chickens, in breeding practice, when the search area is narrowed down to a few square meters (corresponding to several grids), staff can accurately identify motionless sick chickens or restless, stressed chickens from this small flock with the naked eye in a very short time. This invention significantly reduces the scope of problem investigation from the entire chicken coop or exercise area (hundreds to thousands of square meters) to a local area with minimal manual verification costs. It avoids the huge computational overhead caused by pixel-level positioning while ensuring that the positioning accuracy is sufficient to meet actual production needs. It achieves a good balance between computational efficiency and practical effect, and significantly improves the practicality and scalability of this invention in large-scale breeding scenarios.

[0080] It should be noted that, in a healthy state, chickens typically move within a certain range, and the activity intensity in each area remains statistically relatively stable. The preferred application scenario for this invention is when cameras cover a fixed exercise area or the entire chicken house, where the chickens' activity area is relatively fixed. For gaps in monitoring that may occur due to normal migration of the chickens in a localized area, the following approach can be used: After the system alarms, staff quickly check the alarmed area on the monitoring screen. If the area is empty (e.g., the chickens have already migrated to another area), the alarm can be ignored without on-site inspection. The system will automatically pause new alarms triggered in that area due to low activity intensity until chickens are detected entering the area and the activity intensity returns to normal. Only then will the complete alarm process be reactivated for that area. This mechanism avoids repeated invalid alarms caused by temporary vacancy of an area and ensures that the monitoring function automatically resumes when the chickens return. If the screen shows chickens present and motionless, further on-site confirmation is required, effectively preventing false alarms caused by empty areas from interfering with on-site management.

[0081] Based on the same application concept, this invention also proposes a chicken flock health monitoring system, comprising: a video image acquisition module for acquiring video images of the chicken flock's activity area; a processor configured to execute the above-described chicken flock health monitoring method; and an alarm module for triggering an alarm when the processor issues a health warning signal.

[0082] The technical solution of this embodiment, through the cooperation of various functional modules, ultimately achieves the monitoring of abnormal individuals in the chicken flock by issuing a health warning signal when the distribution characteristic parameters deviate abnormally from the baseline parameters. Furthermore, this invention can calculate the spatial distribution data of the chicken flock's movement by comparing images, requiring minimal computation, no high-performance GPU servers, no individual data, and no sensors on the chickens, thus reducing costs and avoiding the stress caused by sensor-equipped chickens. Moreover, this invention monitors the entire chicken flock, completely independent of target detection and multi-target tracking technologies, avoiding tracking failures in scenarios with high stocking density and chickens obscuring each other. This invention uses cameras to collect video images, providing 24-hour coverage, and calculates real-time monitoring data for the chicken flock. Abnormal deviations in the monitoring data trigger real-time health warning signals, preventing detection only when obvious symptoms appear, ensuring high timeliness and avoiding subjective judgment differences inherent in manual inspections.

[0083] Furthermore, this invention also proposes a computer-readable storage medium, characterized in that a chicken health monitoring program is stored on the computer-readable storage medium, and when the chicken health monitoring program is executed by a processor, it implements the chicken health monitoring method as described above. Therefore, it will not be described again here. In addition, the beneficial effects of using the same method will not be described again. For technical details not disclosed in the embodiments of the computer-readable storage medium involved in this invention, please refer to the description of the method embodiments of this invention. The program instructions can be deployed to be executed on a computing device, or on multiple computing devices located in one location, or on multiple computing devices distributed in multiple locations and interconnected through a communication network. Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The above-mentioned program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above methods. The above-mentioned storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0084] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without creative effort. Through the above description of the embodiments, those skilled in the art can clearly understand that this invention can be implemented by means of software plus necessary general-purpose hardware, and of course, it can also be implemented by dedicated hardware including dedicated integrated circuits, dedicated CPUs, dedicated memory, dedicated components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can be diverse, such as analog circuits, digital circuits, or dedicated circuits. However, for this invention, software program implementation is more often a preferred implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, portable hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0085] The above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for monitoring the health of chicken flocks, characterized in that, include: Collect video images of the chickens' activity area; Based on the video images, distribution data reflecting the spatial distribution of the chicken flock's movement intensity is generated; Based on the distribution data, distribution feature parameters are extracted to describe the spatial distribution characteristics of the chicken flock's movement intensity; Compare the current distribution characteristic parameters with the corresponding baseline parameters established under historical health conditions; When the comparison results indicate that the current distribution characteristic parameter deviates abnormally from the baseline parameter, a health warning signal is issued.

2. The method for monitoring chicken flock health as described in claim 1, characterized in that, The distribution characteristic parameter includes kurtosis; the abnormal deviation includes the increase in kurtosis relative to the baseline parameter kurtosis baseline exceeding a predetermined threshold.

3. The method for monitoring chicken flock health as described in claim 2, characterized in that, The distribution characteristic parameters also include skewness; the health warning signal includes a hybrid abnormal signal, which is issued when the skewness does not exceed the preset range of the skewness baseline in the baseline parameters but the kurtosis deviates abnormally.

4. The method for monitoring chicken flock health as described in claim 1, characterized in that, The baseline parameters are established separately for multiple pre-divided time periods throughout the day.

5. The method for monitoring chicken flock health as described in claim 4, characterized in that, The baseline parameters are dynamically updated using a sliding time window or an exponentially weighted moving average.

6. The method for monitoring chicken flock health as described in any one of claims 1 to 5, characterized in that, The step of generating distribution data reflecting the spatial distribution of chicken movement intensity based on the video image includes: The motion information of the chicken flock is generated based on the video images; wherein, the motion information is obtained by the inter-frame difference method: the pixel grayscale difference between two adjacent frames or two frames with a predetermined interval in the video image sequence is calculated as the motion information; The distribution data is generated based on the motion information.

7. The method for monitoring chicken flock health as described in any one of claims 1 to 5, characterized in that, The step of generating distribution data reflecting the spatial distribution of chicken movement intensity based on the video image includes: The video image is divided into multiple spatial grids, and the statistical value of motion intensity within each grid is calculated. The set of these statistical values ​​is used as the distribution data.

8. The method for monitoring chicken flock health as described in any one of claims 1 to 5, characterized in that, Also includes: The system identifies low-motion regions where the motion intensity is below a predetermined threshold, tracks the displacement of the center of mass of these regions over a continuous period of time, and issues a warning signal when the displacement is less than the predetermined threshold and the duration exceeds a predetermined value.

9. A chicken flock health monitoring system, characterized in that, include: The video image acquisition module is used to acquire video images of the area where the chickens are active. A processor configured to perform the flock health monitoring method according to any one of claims 1 to 8; The alarm module is used to issue an alarm when the processor sends a health warning signal.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a flock health monitoring program, which, when executed by a processor, implements the flock health monitoring method as described in any one of claims 1 to 8.

Images