Farm chicken health monitoring method and system

Abstract

The invention provides a livestock farm chicken health monitoring method, which comprises the following steps: S1, acquiring all frames in a corresponding video clip in an activity area in which the change rate of the number of chickens is smaller than a set threshold value K by using a posture recognition algorithm, and outputting five articulation points of chicken bones of all the frames, including a parietal bone, a clavicle, a left metatarsal bone, a right metatarsal bone and a caudal vertebra; and S2, judging whether the chicken is in a curled state or not according to the information of the five joint points of the chicken skeleton, if so, judging that the health of the chicken flock is abnormal and carrying out early warning, and otherwise, entering the step S1. According to the method, the activity areas of the chickens are monitored, when the activity degree of the chickens in a certain activity area is too low, the bones of the chickens are obtained through a posture recognition algorithm, then whether the chickens suffer from diseases or not is judged according to the position relation of the bones of the chickens, unattended automatic recognition and monitoring of chicken flock health are achieved, and the chicken flock health monitoring efficiency is improved. And the efficiency and instantaneity are greatly improved.

Description

Technical Field

[0001] The invention relates to the field of chicken breeding methods, and in particular to a method and system for monitoring the health of chickens in a breeding farm. Background Art

[0002] Common diseases in chickens include Marek's disease, Newcastle disease, avian influenza, infectious bursal disease, etc. The above diseases can cause abnormal postures in chickens. In the prior art, it is generally determined whether a chicken is sick by manually observing the movement state of the chicken and whether the chicken has abnormal posture. This method is very time-consuming and there is no system or method in the prior art to determine whether a chicken is sick based on the chicken's posture. Summary of the invention

[0003] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a method and system for monitoring the health of chickens in a farm to solve the problems mentioned in the above background technology section.

[0004] The present invention is achieved through the following technical solutions:

[0005] A method for monitoring the health of chickens in a breeding farm, comprising:

[0006] S1, using the posture recognition algorithm, obtain all frames in the corresponding video clip in the activity area where the change rate of the number of chickens is less than the set threshold K, and output the five joint points of the chicken bones in all frames, including the parietal bone, clavicle, left metatarsal bone, right metatarsal bone, and coccyx, wherein the five joint points of the chicken bones are obtained by the following formula:

[0007]

[0008] In the formula, C i represents the coordinates of each joint point, M represents the number of feature points corresponding to the joint point, α ij represents the influence weight of feature point j on each joint point i, F j represents the position of feature point j;

[0009] S2, judging whether the chicken is in a "curled up state" according to the information of the five joints of the chicken skeleton, if yes, judging that the chickens are in a "curled up state" and giving an early warning, otherwise proceeding to step S1; wherein, in step S2, judging whether the chicken is in a "curled up state" according to the information of the five joints of the chicken skeleton is obtained by the following formula:

[0010] Δy=y top -y tail

[0011] Where Δy represents the y coordinate difference between the parietal joint and the coccygeal joint, y toprepresents the y coordinate of the parietal joint point, y tail represents the y coordinate of the coccygeal joint point;

[0012] When Δy is less than the set threshold, it is judged that the chicken is in a curled-up state.

[0013] The beneficial effects of the present invention are as follows: the present invention monitors the activity areas of the chickens. When the activity of the chickens in a certain activity area is too low, the posture recognition algorithm is used to obtain the joint point information of the chicken bones, and then the positional relationship of the chicken bones is used to determine whether the chickens are sick, thereby realizing unattended automatic identification and monitoring of the health of the chicken flock, greatly improving efficiency and immediacy. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The present invention provides a flow chart of a method for monitoring the health of chickens in a breeding farm.

[0015] Figure 2 The figure is a schematic diagram of the positions of five joints of the chicken skeleton of the present invention, namely, the parietal bone, clavicle, left metatarsal bone, right metatarsal bone and coccyx. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention. In addition, the technical features involved in each embodiment of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0017] Reference Figure 1 to Figure 2 As shown, a method for monitoring chicken health in a farm comprises:

[0018] S1, using the posture recognition algorithm, obtain all frames in the corresponding video clip in the activity area where the change rate of the number of chickens is less than the set threshold K, and output the five joints of the chicken bones in all frames, including the parietal bone, clavicle, left metatarsal bone, right metatarsal bone, and coccyx. Figure 2 As shown, the parietal bone is at position 1, the clavicle is at position 2, the left metatarsal is at position 3, the right metatarsal is at position 4, and the coccyx is at position 5.

[0019] S2, judging whether the chicken is in a "curled up state" based on the information of the five joints of the chicken's skeleton, if so, determining that the health of the chickens is abnormal and issuing an early warning, otherwise, proceeding to step S1.

[0020] Among them, in the step S1, the step of obtaining all frames of the corresponding video clip in the activity area where the change rate of the number of chickens is less than the set threshold K specifically includes:

[0021] S101, dividing the activity areas of the chicken house into A1, A2, ..., Ax, and performing video surveillance on each activity area;

[0022] S102, obtaining the change rate K of the chickens in each activity area at a predetermined interval AX ;

[0023] S103, select some chickens from the flock as sample N, and after a predetermined time interval, count the number of chickens n in the sample N whose activity areas have changed, and the sample change rate K 样 =n / N; wherein, taking some chickens in the flock as samples N is achieved by setting displacement sensors on the chickens; the displacement sensors are such as GPS trackers, radio telemetry trackers, etc.

[0024] S104, set K AX The corresponding weight values ​​λ1, K 样 The corresponding weight value λ2, then the chicken group change rate K corresponding to a certain activity area X =λ1*K 样 +λ2*K AX , when K X When it is less than the set threshold K, the process goes to step S2.

[0025] In step S102, K AX Calculated by the following formula:

[0026]

[0027] Where Axy represents the number of chickens counted in chronological order in the activity area Ax, and is recorded as Ax1, Ax2, ..., Axy respectively.

[0028] Through the above steps, a small number of chickens are selected as sample chickens, and the corresponding weight values ​​are set according to the change rate of the sample chickens and the change rate of the number of chickens in each active area in the entire chicken flock, so as to realize the calculation of the overall activity of the chicken flock. The sample chickens can be tracked by bundling with trackers, thereby realizing the selection of a small number of chickens as sample chickens and cooperating with video monitoring to realize accurate judgment of the activity of the entire chicken flock. When the activity of the chicken flock in a certain area is too low, the health of the chicken flock can be determined by step S2, which can reduce the amount of calculation of the present invention.

[0029] For example: Assume that the total number of chickens raised is 500, and 50 chickens are selected as sample chickens N. Assume that the activity areas of 10 chickens n have changed, then K2 = 10 / 50 = 0.2. Assume that the number of chickens before and after video surveillance in area A1 is 70 and 100, then K A1 =30 / 100=0.3. In one embodiment, in step S104, the weight value λ1 ranges from 0.5 to 0.8.

[0030] Among them, through experiments, when the weight values ​​λ1 and λ2 are taken as follows respectively, the accuracy ratio of the chicken activity counted by the present invention to the manual count is as shown in the following table.

[0031] <![CDATA[λ1]]> 0.5 0.55 0.6 0.65 0.7 0.75 0.8 <![CDATA[λ2]]> 0.5 0.45 0.4 0.35 0.3 0.25 0.2 Accuracy 70％ 74％ 78％ 83％ 75％ 72％ 70％

[0032] In step S1, the five joint points of the chicken skeleton are obtained by the following formula:

[0033]

[0034] In the formula, C i represents the coordinates of each joint point, M represents the number of feature points corresponding to the joint point, α ij represents the influence weight of feature point j on each joint point i, F j represents the position of feature point j.

[0035] The specific steps can refer to the following embodiments.

[0036] In the input image I, the feature map is first extracted through a convolutional neural network (CNN), assuming that the size of the feature map is 64x64.

[0037] Using the information in the feature map, the parietal joint point P is calculated through a fully connected layer k Assume that P k The coordinates are (30,45).

[0038] Define a weight matrix W k , where each element W ij Represents the influence weight of feature point j on parietal joint point i.

[0039] Assume that the number of feature points M is 10 and the weight matrix W k The size is 10x10.

[0040] Calculate the position F of each feature point j j , assuming that the coordinates of feature point j are (20,35).

[0041] By formula C i =Σ(α ij *Fj ), where α ij Represents the influence weight of feature point j on parietal joint point i, and calculates the coordinates C of the parietal joint point i .

[0042] Assume α ij The value of C is 1, i =1*(20,35)=(2,5).

[0043] In this way, the coordinates of all parietal joints can be obtained.

[0044] Finally, these coordinates are compared with the input image I to verify the accuracy of the calculation results.

[0045] The whole process can be implemented by writing Python code, using the NumPy library for matrix operations and the OpenCV library for image processing.

[0046] Through this method, the precise location of the parietal joints can be achieved, and basic data can be provided for subsequent bone analysis.

[0047] And according to the above method, 5 joint points of the chicken skeleton are obtained, which include the parietal bone, clavicle, left metatarsal bone, right metatarsal bone, and coccygeal vertebra.

[0048] In step S2, the information of the five joints of the chicken skeleton is used to determine whether the chicken is in a "curled state", which is obtained by the following formula:

[0049] Δy=y top -y tail

[0050] Where Δy represents the y coordinate difference between the parietal joint and the coccygeal joint, y top represents the y coordinate of the parietal joint point, y tail represents the y coordinate of the coccygeal joint point;

[0051] When Δy is less than the set threshold, it indicates that the parietal joint and the coccyx are too close, and the chicken is judged to be in a "curled state". The set threshold is related to the monitored chicken age, chicken breed, etc.

[0052] Furthermore, by calculating the ratio of the number of frames in which the chickens are in a "curled up state" to the total number of frames in the acquired video, when the ratio exceeds a set value, for example, more than 70% of the time, the chickens are in a "curled up state", the alarm level of the abnormal health warning of the chicken flock is further increased.

[0053] Common infectious diseases in chickens such as enterotoxin syndrome, Newcastle disease, E. coli or salmonella infection can easily cause chickens to be listless, unwilling to move, paralyzed in both legs, lying still, and staying in a "curled up state" for a long time. Therefore, when chickens are in a "curled up state" for a long time, it indicates that the chickens may have the above health problems. In order to avoid the above sick chickens infecting the entire flock, farmers should be reminded to pay attention to the health of the flock.

[0054] Step S2 also includes step S3,

[0055] S3. Use an image segmentation algorithm to extract the comb and wattle areas of the chicken in the "curled up state" from the video frame, and then use a color recognition algorithm to determine whether the color of the comb and wattle of the chicken is abnormal. If the color of the comb or wattle is abnormal, such as white or purple, the alarm level of abnormal health of the chicken flock is further increased.

[0056] Specifically, the RGB color values ​​of the comb and wattle regions can be obtained based on the acquired image data, and the color feature information of the comb can be obtained through these values. The above color values ​​are converted into the HSV color space to obtain hue, saturation and brightness parameters that are more suitable for analysis, and determine whether the color of the comb and wattle is within the healthy range.

[0057] As an extended embodiment, in step S2, based on the information of the five joint points of the chicken's skeleton, it can also be determined whether the chicken is in a "head abnormal trembling state" or a "chicken claw split state". If so, the health of the chicken flock is determined to be abnormal and an early warning is issued, otherwise, step S1 is entered.

[0058] The step of judging whether the chicken is in a "state of abnormal head trembling" according to the information of the five joint points of the chicken skeleton specifically includes:

[0059] When acquiring the first frame, the coordinate system connection line between the parietal joint point and the clavicle joint point;

[0060] When acquiring the second frame, the coordinate system connection line between the parietal joint point and the clavicle joint point;

[0061] Calculate the swing angle θ1 of the line connecting the two coordinate systems in the first frame and the second frame. When the swing θ1 is greater than the set value, count it once, and calculate the swing angle θ of the line connecting the coordinate systems of adjacent frames in the video clip by analogy. n When the total number of times counted exceeds the set value, it is judged that the chicken is in an "abnormal head trembling state".

[0062] Among them, the swing angle θ1 is obtained by the following formula:

[0063]

[0064] In the formula, the coordinates of the chicken parietal joint point in the first frame are P1 = (x1, y1), the coordinates of the clavicle joint point are Q1 = (x2, y2), the coordinates of the chicken parietal joint point in the second frame are P2 = (x3,, y3), and the coordinates of the clavicle joint point are Q1 = (x4, y4).

[0065] In one embodiment, assuming that each acquired video clip is 10 to 20 seconds long, 24 frames are generated per second, and the swing angle θ is set n If the swing angle is greater than the setting θ, it will be recorded once. θ is set to 5°. In each second of video, assuming the swing angle θ n If the number of times greater than θ exceeds 12 times, the chicken's swinging frequency is abnormal, which is why it is determined that the chicken has an "abnormal head shaking state."

[0066] Symptoms of common infectious diseases in chickens, such as Newcastle disease and Mycoplasma synoviae, include neurological symptoms such as head tremors and twisted necks. Therefore, when a chicken has an "abnormal head tremor", it indicates that the chicken may have the above health problems. In order to avoid the above sick chickens infecting the entire flock, farmers should be reminded to pay attention to the health of the flock.

[0067] The step of judging whether the chicken is in the "chicken claw split state" according to the information of the five joint points of the chicken skeleton specifically includes:

[0068] The distance d between the left metatarsal joint and the coccygeal joint is obtained by the following formula: left_foot

[0069]

[0070] In the formula, x left_foot and left_foot Represent the x and y coordinates of the left metatarsal joint, respectively. tail and tai Respectively represent the x and y coordinates of the coccygeal joint;

[0071] The distance d between the right metatarsal joint and the coccygeal joint is obtained by the following formula: right_foot ,

[0072]

[0073] In the formula, x right_foot and right_foot Represent the x and y coordinates of the right metatarsal joint, respectively. tail and tail Respectively represent the x and y coordinates of the coccygeal joint;

[0074] When left_foot and d right_footWhen at least one value exceeds the set threshold, that is, the distance from the left metatarsal joint to the coccygeal joint or the right metatarsal joint to the coccygeal joint is greater than the set threshold, the chicken is in the split state, and the chicken is judged to be in the "split state". The set threshold is related to the monitored chicken age, chicken breed, etc.

[0075] Symptoms of common infectious diseases in chickens, such as Marek's disease, encephalomyelitis, and mycoplasma synoviae, include loss of control of the sciatic nerve and brachial nerve, which leads to movement disorders and paralysis in sick chickens. The "splits" posture is a typical symptom of sciatic nerve damage. Therefore, when chickens have "splits", it indicates that the chickens may have the above health problems. In order to avoid the above sick chickens infecting the entire flock, farmers are reminded to pay attention to the health of the flock.

[0076] In another aspect, the present invention provides a chicken health monitoring system for a breeding farm, comprising:

[0077] The video module sets a predetermined interval and acquires video of the chicken's activity area; wherein the video module may be an image sensor, a camera, etc. The predetermined interval may be acquired during the period of feeding the chickens, or during the chickens' active time such as early morning and evening. Therefore, by performing video monitoring during the above time periods, it may be easy to determine and find out whether there are sick chickens with abnormal health in the chickens.

[0078] The first judgment module is used to process the video obtained by the video module and judge whether the change rate of the number of chickens in the activity area is less than a set threshold, and if so, transmit the obtained video to the processing module;

[0079] A processing module is used to receive the video image sent by the judgment module, and output five joint points of the chicken bones in all frames using a posture recognition algorithm, including the parietal bone, clavicle, left metatarsal bone, right metatarsal bone, and coccyx;

[0080] The second judgment module is used to calculate the positional relationship between the parietal joint and the clavicle joint. When the positional relationship is less than a set threshold, it is judged that the chicken is in a "curled state" and an early warning is issued for the health of the chicken flock.

[0081] In the description of the present invention, it is to be understood that the terms “center”, “longitudinal”, “lateral”, “length”, “width”, “thickness”, “up”, “down”, “front”, “back”, “left”, “right”, “vertical”, “horizontal”, “top”, “bottom”, “inside”, “outside”, “clockwise”, “counterclockwise”, “axial”, “radial”, “circumferential”, etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0082] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0083] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for monitoring the health of chickens in a farm, characterized in that: include:: S1, using the posture recognition algorithm, obtain all frames in the corresponding video clip in the activity area where the change rate of the number of chickens is less than the set threshold K, and output the five joint points of the chicken bones in all frames, including the parietal bone, clavicle, left metatarsal bone, right metatarsal bone, and coccyx, wherein the five joint points of the chicken bones are obtained by the following formula: In the formula, C i represents the coordinates of each joint point, M represents the number of feature points corresponding to the joint point, α ij represents the influence weight of feature point j on each joint point i, F j represents the position of feature point j; S2, judging whether the chicken is in a "curled up state" according to the information of the five joints of the chicken skeleton, if yes, judging that the chickens are in a "curled up state" and giving an early warning, otherwise proceeding to step S1; wherein, in step S2, judging whether the chicken is in a "curled up state" according to the information of the five joints of the chicken skeleton is obtained by the following formula: Δy=y top -y tail Where Δy represents the y coordinate difference between the parietal joint and the coccygeal joint, y top represents the y coordinate of the parietal joint point, y tail represents the y coordinate of the coccygeal joint point; When Δy is less than the set threshold, the chicken is judged to be in a "curled state".

2. A chicken health monitoring method for a farm according to claim 1, characterized in that: In step S1, the step of obtaining all frames in the corresponding video clip in the activity area where the change rate of the number of chickens is less than the set threshold K specifically includes: S101, dividing the activity areas of the chicken house into A1, A2, ..., Ax, and performing video surveillance on each activity area; S102, obtaining the change rate K of the chickens in each activity area at a predetermined interval AX ; S103, select some chickens from the flock as sample N, and after a predetermined time interval, count the number of chickens n in the sample N whose activity areas have changed, and the sample change rate K 样 =n / N, wherein the taking of some chickens in the flock as samples N is achieved by setting displacement sensors on the chickens; S104, set K AX The corresponding weight values ​​λ1, K 样 The corresponding weight value λ2, then the chicken group change rate K corresponding to a certain activity area X =λ1*K 样 +λ2*K AX , when K X When it is less than the set threshold K, the process goes to step S2.

3. A chicken health monitoring method for a farm according to claim 2, characterized in that: In step S102, K AX Calculated by the following formula: Where Axy represents the number of chickens counted in chronological order in the activity area Ax, and is recorded as Ax1, Ax2, ..., Axy respectively.

4. A chicken health monitoring method for a farm according to claim 2, characterized in that: In step S104, the weight value λ1 ranges from 0.5 to 0.

8.

5. The chicken health monitoring method of claim 1, wherein: Step S2 also includes step S3, S3. Use an image segmentation algorithm to extract the comb and wattle areas from the video frame, and then use a color recognition algorithm to determine whether the color of the comb and wattle is abnormal. If the color of the comb or wattle is abnormal, the alarm level of abnormal health of the chicken flock is further increased.

6. A chicken health monitoring system for a breeding farm, characterized in that: include: Video module, set the interval and predetermined time, and obtain the video of the chicken's activity area; The first judgment module is used to process the video obtained by the video module and judge whether the change rate of the number of chickens in the activity area is less than a set threshold, and if so, transmit the obtained video to the processing module; A processing module is used to receive the video image sent by the judgment module, and output five joint points of the chicken bones in all frames using a posture recognition algorithm, including the parietal bone, clavicle, left metatarsal bone, right metatarsal bone, and coccyx; The second judgment module is used to calculate the positional relationship between the parietal joint and the clavicle joint. When the positional relationship is less than a set threshold, it is judged that the chicken is in a "curled state" and an early warning is issued for the health of the chicken flock.

Images