A weighing measurement method for livestock and poultry breeding and its system and medium

Abstract

The present invention provides a weighing and measuring method for livestock and poultry breeding, and a system and medium thereof. When a flexible film sensor has pressure data, a livestock and poultry ID number, posture sensor data and position coordinates corresponding to the pressure data are obtained through a position identifier, and dynamic feature images of livestock and poultry are photographed from different positions through a camera. The position coordinates corresponding to the pressure data are obtained through coordinate conversion, and then the pressure data, the livestock and poultry ID number, posture data and the corresponding dynamic feature head portrait of the livestock and poultry are obtained through the position coordinates. Finally, the pressure data, posture data and dynamic feature image data of the flexible film sensor are used as input parameters, and the actual weight of the livestock and poultry is used as an output parameter, so that a non-structural equation can be constructed. The pig weighing model obtained after fitting is used to measure the weight of the livestock and poultry to be measured, which is convenient for weighing and subsequent breeding.

Description

Technical Field

[0001] The invention relates to the technical field of livestock and poultry breeding, and in particular to a weighing and measuring method for livestock and poultry breeding and a system and a medium thereof. Background Art

[0002] With today's technological innovations, modern factory-based intensive livestock and poultry farming has made great progress. In the breeding process of pigs, sheep and other livestock and poultry, it is often necessary to conduct non-contact measurement and monitoring of their weight, which is not only related to the improvement of feeding efficiency, but also closely related to disease prevention and control. At the same time, it can monitor the health status of livestock and poultry without stress, thereby facilitating the subsequent adjustment of feeding plans.

[0003] In livestock and poultry farming, taking pigs as an example, the existing technologies for measuring pig weight include direct measurement and contactless indirect measurement. Direct measurement is a method of manually driving pigs into weighing equipment to measure their weight. However, errors are inevitable during manual operation, affecting the accuracy and reliability of the data. Contactless indirect measurement is a method of contactless identification and detection of pigs. The images of pigs are obtained through cameras, and the weight of the pigs is estimated based on image recognition technology.

[0004] For example, the Chinese patent application No. 201910153717.2, with a publication date of September 8, 2020, discloses a method and device for estimating a target weight, which relates to the field of measurement. The method includes: extracting the characteristic value of the target based on the body size information of the target to be estimated, wherein a plurality of characteristic values ​​constitute a characteristic value set; determining the maximum weight value and the minimum weight value corresponding to the characteristic value set according to the correspondence between the characteristic value of the sample target and the weight range; determining the number of characteristic values ​​in the characteristic value set corresponding to each weight value between the maximum weight value and the minimum weight value; and taking the average value of the weight values ​​corresponding to the characteristic values ​​exceeding a predetermined number as the weight value of the target to be estimated. The present disclosure can improve the accuracy of target weight prediction.

[0005] The above-mentioned document acquires the image of the pig and then estimates the weight based on the image information of the pig. However, it only estimates the weight based on the image information. There may be multiple pigs in the image of the pig. The current posture of the pig, such as looking down while eating, will cause the center of gravity of the pig to drop, which will affect the weight of the pig, resulting in unreliable pig weighing. Summary of the invention

[0006] The object of the present invention is to provide a weighing measurement method and system and medium for livestock and poultry breeding, which can detect the actual weight of livestock and poultry in real time, realize dynamic monitoring of the weight of livestock and poultry to be measured, and facilitate subsequent breeding.

[0007] To achieve the above object, the present invention provides a weighing and measuring method for livestock and poultry breeding, comprising the following steps:

[0008] S1 sets an ID number, identification tag and posture sensor on each livestock and poultry, and a position identifier is installed in the test site. The position identifier is used to obtain the ID number, position coordinates and posture data of each livestock and poultry in the test site;

[0009] S2 collects pressure data when livestock and poultry are in contact with a pressure sensor set on the ground of the breeding ground. During the period when pressure data is displayed, the position coordinates of the identification tag when pressure data is displayed are obtained through the position identifier, and the livestock and poultry ID number, pressure data and posture data corresponding to the position coordinates of the identification tag on the pressure sensor at this time are obtained. At the same time, the dynamic feature image of the livestock and poultry is taken in real time through the camera;

[0010] S3 performs instance segmentation on the dynamic feature image in S2 through an instance segmentation algorithm to obtain an image instance of livestock and poultry, and then converts the pixel coordinates of the image instance of livestock and poultry into the position coordinates in S2 through coordinate conversion, and obtains the dynamic feature image of livestock and poultry active on the pressure sensor corresponding to the pressure data and posture data at this time according to the position coordinates, and saves the ID number, pressure data, posture data and dynamic feature image of the livestock and poultry active on the pressure sensor at this time into the database of the weighing system;

[0011] S4 uses the pressure data, posture data and dynamic feature images of livestock and poultry in the weighing system database as input parameters and the actual weight of livestock and poultry as output parameters to construct a non-structural equation, and fits the non-structural equation through keras deep learning to obtain a livestock and poultry weighing model;

[0012] S5 inputs the ID number, posture data, dynamic feature image data and pressure data of the livestock and poultry to be tested into the weighing system, and the actual weight of the livestock and poultry to be tested can be obtained through the weighing model output.

[0013] The above setting sets an ID number on each livestock and poultry, and then measures the weight of livestock and poultry with different ID numbers entering the weighing area through a pressure sensor, and obtains the position coordinates of the identification tag when there is pressure data through a position identifier, thereby obtaining the livestock and poultry ID number, pressure data and posture data corresponding to the position coordinates of the identification tag; by converting the pixel coordinates of the dynamic feature image into the position coordinates of the identification tag when there is pressure data, the dynamic feature image of the livestock and poultry active on the pressure sensor corresponding to the pressure data and posture data at this time is obtained according to the position coordinates, and then the final weight data of the corresponding ID number is determined according to the posture data, the dynamic feature image and the pressure data, so that the posture data and the dynamic feature image data can be introduced in the learning process of the model to determine the final weighing model. In the subsequent weighing process, only the ID number and other data need to be introduced to determine the weight data of the pig, and the weighing method is simple, and the weighing takes into account the behavioral characteristics reflected by the posture data, dynamic image characteristics, etc., so that the weight measurement method is more accurate, which is convenient for adjusting the subsequent feeding plan according to the weight data.

[0014] Furthermore, in step S1, the position identifier is an RFID identifier, the position identification tag is an RFID tag, the position identifier is installed in the farm, and the position identification tag is fixed on the ear of the livestock and poultry; the attitude sensor is an acceleration and gyroscope sensor, and the attitude sensor is fixed on the back of the neck of the livestock and poultry.

[0015] The above settings, without affecting the normal activities of livestock and poultry, make it easy for the RFID identifier in the farm to read the ID number and local location data of livestock and poultry, and identify the posture data through the posture sensor.

[0016] Furthermore, in step S4, a mean filter algorithm is used to perform denoising and burr removal preprocessing on the collected pressure data of the pressure sensor.

[0017] The above settings can reduce the burr interference on the pressure data during the dynamic characteristics of the pigs.

[0018] Furthermore, in step S2, the target detection algorithm is used to enable more than two cameras to capture the dynamic characteristics of livestock and poultry from different positions in the range of the pressure sensor in real time at different angles, and then map them onto the same livestock and poultry to obtain dynamic characteristic images of the livestock and poultry. The dynamic characteristics of the livestock and poultry include the current action behavior of the livestock and poultry and the body parts of the livestock and poultry corresponding to the dynamic characteristics.

[0019] The above arrangement can improve the ability to obtain different dynamic feature images of livestock and poultry through image detection, thereby determining the current movements of the livestock and poultry and the corresponding parts.

[0020] Further, in step S4, the posture data includes acceleration data and body posture inclination data of the dynamic characteristics of livestock and poultry, and the non-structural equation is as follows:

[0021] ;

[0022] in The polynomial equations are divided into acceleration data corresponding to the livestock ID number, body posture inclination data, pressure data of the pressure sensor, and dynamic feature image data. is the actual weight of the livestock and poultry to be tested, is the weight matrix, the weight matrix It is obtained through BP neural network training.

[0023] With the above settings, after measuring the pressure data of the pressure sensor, the posture data of the posture sensor and the dynamic feature image data of the livestock and poultry, the corresponding weight matrix can be obtained through non-structural equation training, and then different output values ​​can be obtained according to different input values, so as to conveniently and accurately calculate the actual weight of the livestock and poultry to be tested.

[0024] Furthermore, in the unstructured equation, To contain The polynomial equation for To contain The polynomial equation for To contain The polynomial equation for To contain The polynomial equation for The calculation formula is as follows:

[0025] ;

[0026] They are The coefficients of the polynomial, The coefficients of the polynomial, The coefficients of the polynomial, The coefficients of the polynomial, where n is a non-negative integer, are obtained through BP neural network training. .

[0027] The above settings enable the BP neural network to calculate The coefficients of the polynomial can then be used to calculate the input parameters of the non-structural equation .

[0028] The present invention also provides a weighing and measuring system for livestock and poultry breeding, comprising a pressure sensor, a position identifier arranged in a breeding farm, a posture sensor, a control module, a database and more than two cameras. The control module is wirelessly connected to the pressure sensor, the position identifier arranged in the breeding farm and the camera respectively through WiFi. The posture sensor is wirelessly connected to the position identifier through WiFi to realize data transmission. When the pressure sensor has pressure data, the position identifier obtains the identification ID number and position coordinates corresponding to the pressure data, and transmits them to the control module to obtain the ID number and posture data corresponding to the position coordinates when there is pressure data. At the same time, the dynamic feature images of livestock and poultry taken from different positions by the camera are transmitted to the control module. The control module is used to perform instance segmentation on the dynamic feature images through an instance segmentation algorithm to obtain image instances of livestock and poultry, and then The pixel coordinates of the livestock and poultry image instances are converted into position coordinates corresponding to the pressure data through coordinate transformation, and the dynamic feature images of the livestock and poultry active on the pressure sensor corresponding to the pressure data and posture data at this time are obtained according to the position coordinates, and the ID number, posture data, dynamic feature image and pressure data of the livestock and poultry corresponding to the position coordinates when there is pressure data are saved in the database; and the pressure data, posture data and dynamic feature image of the livestock and poultry are used as input parameters, and the actual weight of the livestock and poultry is used as output parameters to construct a non-structural equation, and the non-structural equation is fitted through keras deep learning to obtain a livestock and poultry weighing model; and the ID number, posture data, dynamic feature image data and pressure data of the livestock and poultry to be tested are input into the weighing system, and the actual weight of the livestock and poultry to be tested can be obtained through the output of the weighing model.

[0029] Furthermore, the pressure sensor is a flexible film sensor, and the pressure sensor is arranged in the feeding area.

[0030] The above arrangement, by placing the pressure sensor in the feeding area, facilitates attracting livestock and poultry to move into the pressure sensor area for weighing.

[0031] Furthermore, the flexible film sensor includes a three-layer structure of a top layer, a middle layer and a bottom layer. The top layer is set as a soft protective film, the middle layer is set as a piezoelectric film, and the bottom layer is set as a sponge layer.

[0032] The above arrangement, by providing a soft protective film, can isolate the direct contact between the pigs and the piezoelectric film, thereby preventing the pigs from damaging the piezoelectric film; the piezoelectric film is arranged to measure the pressure applied by the pigs through deformation; and the sponge layer is arranged to provide a deformation space for the piezoelectric film.

[0033] The present invention further provides a storage medium for storing a computer program, wherein when the computer program is executed by a processor, the weighing and measuring method for livestock and poultry breeding as described above is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram of the framework of the present invention.

[0035] Figure 2 It is a flow chart of the working method of the present invention.

[0036] Figure 3 It is a schematic diagram of the non-structural equation in the present invention. DETAILED DESCRIPTION

[0037] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] Example 1.

[0039] like Figure 1-3 As shown, the present invention provides a simple wireless weighing and measuring method for livestock and poultry, and the specific steps are as follows:

[0040] S1 sets an ID number, an identification tag and a posture sensor on each livestock and poultry. A position identifier is installed in the test site. The position identifier is used to obtain the ID number, the position coordinates and posture data of each livestock and poultry in the test site. In this embodiment, the livestock and poultry take pigs as an example. The identification tag 1 is fixed on the ear of the pig. The identification tag 1 is connected to the position identifier by wireless so that the position identifier can obtain the position coordinates of the identification tag 1. The position identifier is set as an RFID identifier, and the identification tag 1 is a wearable tag with a built-in RFID chip;

[0041] S2 collects pressure data when livestock and poultry come into contact with a pressure sensor set on the ground of the breeding ground. In this embodiment, the posture sensor 2 is fixed on the back of the pig's neck. The posture sensor 2 is set as an acceleration and gyroscope sensor. The posture data of the pig can be measured by the acceleration and gyroscope sensor. The pressure sensor is a flexible film sensor. The flexible film sensor is set on the four sides of the feeding trough in the pig feeding area. When the pig moves on the four sides of the feeding trough, the flexible film sensor records the pressure data of the pig's stepping. During the time period when the pressure data is displayed, the position coordinates of the identification tag on the pig when there is pressure data and the posture data of the pig at this time are obtained through the position identifier. , obtain the pig ID number, pressure data and posture data corresponding to the position coordinates of the identification tag on the pressure sensor at this time, and at the same time, through the target detection algorithm, make more than two cameras 3 take real-time photos of the dynamic characteristics of the pig from different positions and angles within the range of the flexible film sensor, and then map them to the same pig to obtain the dynamic characteristic image of the pig. In this embodiment, the dynamic characteristics of the pig include the pig's eating, sleeping, excretion, fighting and other actions and the pig's body parts corresponding to the dynamic characteristics, and the body parts include the pig's left front foot, left hind foot, right front foot, right hind foot, etc., wherein the target detection algorithm is a prior art and will not be repeated here;

[0042] S3 performs instance segmentation on the dynamic feature image in S2 through an instance segmentation algorithm to obtain an image instance of a pig, then converts the pixel coordinates of the image instance of the pig into the position coordinates in S2 through coordinate conversion, obtains the dynamic feature image of the pig active on the pressure sensor corresponding to the pressure data and the posture data at this time according to the position coordinates, and saves the ID number, pressure data, posture data and dynamic feature image of the pig active on the pressure sensor at this time into the database of the weighing system. In this embodiment, the instance segmentation algorithm is a prior art and will not be described in detail here;

[0043] S4 uses a mean filtering algorithm to perform noise and burr removal preprocessing on the collected flexible film sensor pressure data, and then uses the pressure data, posture data and dynamic feature images of the pigs in the database of the weighing system as input parameters, and uses the actual weight of the pigs as the output parameter to construct a non-structural equation. In this embodiment, the data of the posture sensor 2 includes the acceleration data and body posture inclination data of the dynamic characteristics of the pigs, and uses the actual weight of the pigs as the output parameter to construct a non-structural equation as follows:

[0044] ;

[0045] In the unstructured equations, is the actual weight of the pig to be tested, is the weight matrix, the weight matrix It is obtained through BP neural network training to test the BP neural network in training weight matrix W The effectiveness and accuracy of the process are evaluated by the BP neural network through the indicators of root mean square error RMSE, mean absolute error MAE and mean absolute percentage error MAPE, so as to obtain the optimal weight matrix in the non-structural equation respectively. In this embodiment, the root mean square error RMSE, the mean absolute error MAE and the mean absolute percentage error MAPE are all prior art and will not be described again;

[0046] To contain The polynomial equation for To contain The polynomial equation for To contain The polynomial equation for To contain The polynomial equation for The calculation formula is as follows:

[0047] ;

[0048] in They are respectively the polynomial of the acceleration data corresponding to the pig ID number, the polynomial of the body posture inclination data, the polynomial of the pressure data of the flexible film sensor and the polynomial of the dynamic feature image data.

[0049] They are The coefficients of the polynomial, The coefficients of the polynomial, The coefficients of the polynomial, The coefficients of the polynomial, where n is a non-negative integer, are obtained through BP neural network training. ;

[0050] The non-structural equation is fitted by keras deep learning to obtain a weighing model. In this embodiment, BP neural network and keras deep learning are both existing technologies and will not be described in detail here.

[0051] S5 inputs the ID number, posture data, dynamic feature image data and pressure data of the pig to be tested into the weighing system, and the actual weight of the pig to be tested can be obtained through the weighing model output.

[0052] For example, when a pig enters the vicinity of a feeding trough and the pressure sensor detects pressure data, the position coordinates of the pig at this time are obtained through the position identifier, and the posture of the pig is detected through the posture sensor, such as the angle information of the pig's body leaning forward, such as leaning forward 30 degrees. The data of the posture sensor can be read simultaneously by the position identifier, and then the dynamic feature image of the pig is obtained through the camera, such as determining that the pig is in a state of eating and that its right foot is on the pressure sensor. This posture, dynamic feature image and pressure data are then recorded and saved in a database. After recording multiple such situations, the pressure data of the pig with this number under this posture and this dynamic feature image is determined through learning, and the final weighing model can be determined through the pig with this number under this posture and this dynamic feature image.

[0053] The working principle of the present invention is as follows: by setting an ID number on each livestock and poultry, then measuring the weight of livestock and poultry with different ID numbers entering the weighing area through a pressure sensor, and obtaining the position coordinates of the identification tag when there is pressure data through a position identifier, thereby obtaining the livestock and poultry ID number, pressure data and posture data corresponding to the position coordinates of the identification tag, and by converting the pixel coordinates of the dynamic feature image into the position coordinates of the identification tag when there is pressure data, thereby obtaining the dynamic feature image of the livestock and poultry active on the pressure sensor corresponding to the pressure data and the posture data at this time according to the position coordinates, and then determining the final weight data of the corresponding ID number according to the posture data, the dynamic feature image and the pressure data, so that the posture data and the dynamic feature image data can be introduced in the learning process of the model to determine the final weighing model, and in the subsequent weighing process, only the ID number and other data need to be introduced to determine the weight data of the pig, and the weighing method is simple, and the weighing takes into account the behavioral characteristics embodied by the posture data, the dynamic image characteristics and the like, so that the weight measurement method is more accurate, and it is convenient to adjust the subsequent feeding plan according to the weight data, such as adjusting the focus feeding when the weight is too light.

[0054] Example 2.

[0055] The present invention also provides a weighing and measuring system for livestock and poultry breeding, comprising a pressure sensor, a position identifier arranged in a breeding farm, a posture sensor 2, a control module, a database and more than two cameras 3, wherein the lens module of the camera 3 comprises a standard lens, an infrared lens and a panoramic lens, so that it is convenient to shoot pigs under normal light conditions through the standard lens, to shoot pigs under night or low light conditions through the infrared lens, and to shoot a wider range of pig images in the breeding farm through the panoramic lens, so as to realize shooting applications in different scenes; in this embodiment, the pressure sensor is a flexible film sensor, and the posture sensor is an acceleration and gyroscope sensor, and the flexible film sensor comprises a three-layer structure of a top layer, a middle layer and a bottom layer, the top layer is set as a soft protective film 11, the middle layer is set as a piezoelectric film 12, and the bottom layer is set as a sponge layer 13, and by setting the soft protective film 11, the direct contact between the pig and the piezoelectric film 12 can be isolated to prevent the pig from damaging the piezoelectric film 12; the piezoelectric film 12 is set to measure the pressure applied by the pig through deformation; and the sponge layer 13 is set to provide a deformation space for the piezoelectric film 12.

[0056] In this embodiment, the control module is wirelessly connected to the flexible film sensor, the position identifier set in the farm and the camera through WiFi respectively. The posture sensor 2 is wirelessly connected to the position identifier through WiFi to realize data transmission. When the flexible film sensor has pressure data, the pig ID number corresponding to the pressure data, the data and position coordinates of the posture sensor 2 are obtained through the position identifier, and transmitted to the control module. At the same time, the dynamic feature images of livestock and poultry are taken from different positions by the camera and transmitted to the control module. The control module is used to perform instance segmentation on the dynamic feature image through an instance segmentation algorithm to obtain image instances of livestock and poultry, and then convert the pixel coordinates of the image instance of livestock and poultry into position coordinates corresponding to the pressure data through coordinate conversion. The mark is obtained according to the position, and the dynamic characteristic images of livestock and poultry active on the pressure sensor corresponding to the pressure data and posture data at this time are obtained, and the ID number, posture data, dynamic characteristic image and pressure data of the livestock and poultry corresponding to the position coordinates when there is pressure data are saved in the database; and the pressure data, posture data and dynamic characteristic image of livestock and poultry are used as input parameters, and the actual weight of livestock and poultry is used as output parameters to construct a non-structural equation, and the non-structural equation is fitted through keras deep learning to obtain a livestock and poultry weighing model; and when the ID number, posture data, dynamic characteristic image data and pressure data of the livestock and poultry to be tested are input into the weighing system, the actual weight of the livestock and poultry to be tested can be obtained through the output of the weighing model.

[0057] Example 3.

[0058] The present invention also provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, a weighing and measuring method for livestock and poultry breeding as described in the above-mentioned embodiment 1 is implemented.

Claims

1. A weighing and measuring method for livestock and poultry breeding, comprising the following steps: S1 sets an ID number, identification tag and posture sensor on each livestock and poultry, and a position identifier is installed in the test site. The position identifier is used to obtain the ID number, position coordinates and posture data of each livestock and poultry in the test site; S2 collects pressure data when livestock and poultry are in contact with a pressure sensor set on the ground of the breeding ground. During the period when pressure data is displayed, the position coordinates of the identification tag when pressure data is displayed are obtained through the position identifier, and the livestock and poultry ID number, pressure data and posture data corresponding to the position coordinates of the identification tag on the pressure sensor at this time are obtained. At the same time, the dynamic feature image of the livestock and poultry is taken in real time through the camera; S3 performs instance segmentation on the dynamic feature image in S2 through an instance segmentation algorithm to obtain an image instance of livestock and poultry, and then converts the pixel coordinates of the image instance of livestock and poultry into the position coordinates in S2 through coordinate conversion, and obtains the dynamic feature image of livestock and poultry active on the pressure sensor corresponding to the pressure data and posture data at this time according to the position coordinates, and saves the ID number, pressure data, posture data and dynamic feature image of the livestock and poultry active on the pressure sensor at this time into the database of the weighing system; S4 uses the pressure data, posture data and dynamic feature images of livestock and poultry in the weighing system database as input parameters and the actual weight of livestock and poultry as output parameters to construct a non-structural equation, and fits the non-structural equation through keras deep learning to obtain a livestock and poultry weighing model; S5 inputs the ID number, posture data, dynamic feature image data and pressure data of the livestock and poultry to be tested into the weighing system, and the actual weight of the livestock and poultry to be tested can be obtained through the weighing model output.

2. A weighing and measuring method for livestock and poultry breeding according to claim 1, characterized in that: In step S1, the position identifier is an RFID identifier, the position identification tag is an RFID tag, the position identifier is installed in the farm, and the position identification tag is fixed on the ear of the livestock; the attitude sensor is an acceleration and gyroscope sensor, and the attitude sensor is fixed on the back of the neck of the livestock.

3. The weighing and measuring method for livestock and poultry breeding according to claim 1, characterized in that: In step S4, the collected pressure data of the pressure sensor is preprocessed by removing noise and burrs using a mean filtering algorithm.

4. The weighing and measuring method for livestock and poultry breeding according to claim 1, characterized in that: In step S2, the target detection algorithm is used to enable two or more cameras to shoot the dynamic characteristics of livestock and poultry from different positions in the range of the pressure sensor in real time at different angles, and then map them onto the same livestock and poultry to obtain the dynamic characteristic images of the livestock and poultry. The dynamic characteristics of the livestock and poultry include the current action behavior of the livestock and poultry and the body parts of the livestock and poultry corresponding to the dynamic characteristics.

5. The weighing and measuring method for livestock and poultry breeding according to claim 1, characterized in that: In step S4, the posture data includes acceleration data and body posture inclination data of the dynamic characteristics of livestock and poultry, and the non-structural equation is as follows: ； in The polynomial equations are divided into acceleration data corresponding to the livestock ID number, body posture inclination data, pressure data of the pressure sensor, and dynamic feature image data. is the actual weight of the livestock and poultry to be tested, is the weight matrix, the weight matrix It is obtained through BP neural network training.

6. A weighing and measuring method for livestock and poultry breeding according to claim 5, characterized in that: In the unstructured equations, To contain The polynomial equation for To contain The polynomial equation for To contain The polynomial equation for To contain The polynomial equation for The calculation formula is as follows: ； They are The coefficients of the polynomial, The coefficients of the polynomial, The coefficients of the polynomial, The coefficients of the polynomial, where n is a non-negative integer, are obtained through BP neural network training. .

7. A weighing and measuring system for livestock and poultry breeding, characterized in that: The invention comprises a pressure sensor, a position identifier arranged in a breeding farm, a posture sensor, a control module, a database and two or more cameras. The control module is wirelessly connected to the pressure sensor, the position identifier arranged in the breeding farm and the camera respectively through WiFi. The posture sensor is wirelessly connected to the position identifier through WiFi to realize data transmission. When the pressure sensor has pressure data, the position identifier obtains the identification ID number and position coordinates corresponding to the pressure data, and transmits them to the control module to obtain the ID number and posture data corresponding to the position coordinates when there is pressure data. At the same time, the dynamic feature images of livestock and poultry taken from different positions by the camera are transmitted to the control module. The control module is used to perform instance segmentation on the dynamic feature images through an instance segmentation algorithm to obtain image instances of livestock and poultry, and then convert the image instances of livestock and poultry into images of livestock and poultry through coordinate conversion. The pixel coordinates of the image instance are converted into the position coordinates corresponding to the pressure data. The dynamic feature images of livestock and poultry active on the pressure sensor corresponding to the pressure data and posture data are obtained according to the position coordinates. The ID number, posture data, dynamic feature image and pressure data of the livestock and poultry corresponding to the position coordinates when there is pressure data are saved in the database. The pressure data, posture data and dynamic feature image of livestock and poultry are used as input parameters, and the actual weight of livestock and poultry is used as output parameters to construct a non-structural equation. The non-structural equation is fitted through keras deep learning to obtain a livestock and poultry weighing model. The ID number, posture data, dynamic feature image data and pressure data of the livestock and poultry to be tested are input into the weighing system, and the actual weight of the livestock and poultry to be tested can be obtained through the output of the weighing model.

8. A livestock and poultry weighing and measuring system according to claim 7, characterized in that: The pressure sensor is a flexible film sensor, and the pressure sensor is arranged in the feeding area.

9. A weighing and measuring system for livestock and poultry breeding according to claim 8, characterized in that: The flexible film sensor comprises a three-layer structure of a top layer, a middle layer and a bottom layer. The top layer is arranged as a soft protective film, the middle layer is arranged as a piezoelectric film, and the bottom layer is arranged as a sponge layer.

10. A storage medium for storing a computer program, characterized in that: When the computer program is executed by a processor, a weighing and measuring method for livestock and poultry breeding as described in any one of claims 1 to 6 is implemented.

Images