System for estimating body size of muscovy duck in living body

Abstract

The invention provides a system for estimating the body size of a muscovy duck in a living body. According to the system, firstly, a body position image of the muscovy duck is collected through a data collection module; then the wireless communication module transmits the body position image to a contour generation module, and the body weight data is transmitted to a data analysis module; after the contour generation module extracts the Muscovy duck edge contour curve, the measurement point marking module marks body size measurement points of the Muscovy duck edge contour curve; the data generation module calculates the obtained body size measurement points and outputs Muscovy duck body size data; and finally, the data analysis module constructs a Muscovy duck growth prediction model for screening Muscovy ducks meeting a preset standard. According to the method, non-contact multi-angle body position images of the Muscovy ducks are acquired, the body size data of the Muscovy ducks are calculated by using the deep learning model, and the growth prediction model is constructed, so that the body size data of the Muscovy ducks are rapidly and accurately predicted, and breeding ducks meeting preset standards are screened; the Muscovy duck body size measuring efficiency is improved, and the stress reaction caused by contact type measurement to the Muscovy duck is avoided.

Description

Technical Field

[0001] The invention relates to the technical field of poultry detection, and in particular to a system for estimating the body size of a living Muscovy duck. Background Art

[0002] In recent years, with the rapid development of modern information technology in my country, more and more automated and intelligent technologies have been developed and widely used in the field of livestock and poultry breeding and production, greatly improving the breeding efficiency of related industries.

[0003] The weight and size of Muscovy ducks often directly or indirectly affect production performance. At present, in order to study the growth and development of Muscovy ducks and the physical characteristics of the breed, in addition to observing the appearance of Muscovy ducks, measuring tools are also used to measure the weight, size and other indicators of Muscovy ducks, analyze the growth and slaughter performance of Muscovy ducks, so as to accumulate basic data for the conservation, breeding, development and utilization of Muscovy ducks.

[0004] In the existing technology, in order to conduct a more comprehensive analysis of the body size indicators of Muscovy ducks, measuring tools such as tape measures, calipers, meter sticks, cotton threads, breast angle meters, and electronic scales are usually used. Although there are many measuring tools and comprehensive measuring indicators, forced measurement during the entire measurement process will cause stress reactions in Muscovy ducks, and manual measurement will lead to inaccurate data of various measurement indicators due to subjective factors; in addition, manual measurement of the weight and body size of Muscovy ducks is a heavy physical labor and consumes a lot of human resources. In view of this, it is necessary to develop a body size measurement system with a simple structure and accurate measurement for Muscovy duck seed conservation and breeding and growth monitoring. Summary of the invention

[0005] The purpose of the present invention is to provide a system for estimating the body size of Muscovy ducks in vivo in order to solve the above problems in the prior art, thereby preventing the ducks from being affected by stress due to contact measurement, accurately measuring and recording their growth performance data, and reducing labor intensity.

[0006] To achieve the above object, the technical solution of the present invention is a system for estimating the body size of a Muscovy duck in vivo, the system comprising:

[0007] A data acquisition module, used to obtain body position images of Muscovy ducks;

[0008] A wireless communication module, used to transmit the body position image acquired by the data acquisition module to the contour generation module, and transmit the weight data to the data analysis module;

[0009] The contour generation module is used to perform semantic segmentation based on the body position image obtained by the data acquisition module using the A-Unet algorithm trained in deep learning to extract the edge contour curve of the Muscovy duck;

[0010] The measurement point marking module is used to mark the body measurement points of the extracted edge contour curve of the Muscovy duck using the dynamic grid method in OpenCV;

[0011] A data generation module is used to calculate and output the body size data of the Muscovy duck according to the ratio of the image pixels in the body image to the pixels of the actual length and the extracted body size measurement points;

[0012] The data analysis module is used to analyze the acquired body size and weight data of Muscovy ducks, predict the growth trajectory of Muscovy ducks and screen breeding ducks based on the analysis results, and output a growth assessment report in the form of a chart.

[0013] Furthermore, the data acquisition module includes a measuring device, which includes a box body, a circular column, an electric control door, a partition and a control main board are provided in the box body, the circular column is arranged at the center of the box body, the control main board is arranged on the circular column, the electric control door is electrically connected to the control main board, the two sides of the electric control door and the partition are respectively connected to the circular column and the box body wall, the partition divides the interior of the box body into multiple measuring units, the electric control door divides the measuring unit into a first space and a second space, and the first space and the second space are respectively provided with door-shaped holes opposite to the circular column in the box body wall.

[0014] Furthermore, the first space includes a depth camera, a distance sensor, a feeding trough, a conveyor belt, a weight sensor and an RFID electronic tag reading module. The conveyor belt is electrically connected to the control mainboard, and the depth camera, the distance sensor, the RFID electronic tag reading module and the weight sensor are respectively connected to the control mainboard signal. The depth cameras are respectively arranged at the top of the first space, the middle of the circular column and the middle of the partition, and are used to collect body position images of the Muscovy duck at different angles. The distance sensor is arranged next to each depth camera, and is used to measure the distance from the depth camera to the Muscovy duck at different angles. The feeding trough is arranged below the depth camera in the middle of the circular column and is detachably connected to the circular column to attract the Muscovy duck to enter the first space. The conveyor belt is separately arranged at the bottom of each measuring unit, and is used to convey the Muscovy duck from the first space to the second space. The weight sensor and the RFID electronic tag reading module are respectively arranged below the conveyor belt of each first space.

[0015] Furthermore, the second space includes a driving mechanism, which includes an electric telescopic rod, a driving part and a mimeograph, the electric telescopic rod is electrically connected to the control mainboard, the mimeograph is arranged on the driving part and is connected to the control mainboard signal, the electric telescopic rod is arranged on a circular column parallel to the ground, and the driving part is detachably connected to the electric telescopic rod, and is used to drive the Muscovy duck out of the second space.

[0016] Furthermore, the specific steps of using the trained A-Unet algorithm to perform semantic segmentation and extract the edge contour curve of the Muscovy duck are:

[0017] S11, processing the original Muscovy duck body position image through the A-Unet model to generate a mask image, performing binarization processing on the mask image, and outputting a binarized image with mask information;

[0018] S12, using the binary image with mask information in step S11, and using the edge finding function findContours to extract the contour curve of the Muscovy duck;

[0019] S13. Use the drawContours function to draw the edge contour curve of the Muscovy duck.

[0020] Furthermore, the steps of marking the body measurement points using the dynamic grid method are as follows:

[0021] S21, using the bounding Rect function in OpenCV to obtain the maximum circumscribed matrix of the edge contour curve of the Muscovy duck from the binary edge contour curve of the Muscovy duck, wherein the maximum circumscribed matrix is ​​a horizontal rectangular rectangle including all key points in the image;

[0022] S22, dividing the image into grid units, and dynamically adjusting the grid units according to feature information of pixels in the units until the pixel coordinates of the edge contour curve of the Muscovy duck are extracted;

[0023] S23, calculating the mean of the x-axis and y-axis coordinates of the pixel points of the edge contour curve of the Muscovy duck, and extracting the corresponding body size measurement points of the Muscovy duck.

[0024] Furthermore, the steps of calculating the body size are:

[0025] S31, calibrating the position of the depth camera, and obtaining a pixel ratio of an image pixel to an actual length at each calibrated position, where the actual length is the distance from the Muscovy duck to the depth camera;

[0026] S32, calculating the pixel length between two body scale measurement points of the image using Euclidean distance calculation;

[0027] S33, converting the pixel length into the actual body size data of the Muscovy duck by using the pixel ratio calculated in step S31 and the pixel length calculated in step S32, and the calculation formula is:

[0028]

[0029] Where L is the body length, (X1, Y1)(X2, Y2) are the coordinates of the measurement points,

[0030] is the pixel length between two measuring points, and P is the pixel ratio.

[0031] Furthermore, the data analysis module includes:

[0032] Data preprocessing module, used to track the weight and size data of Muscovy ducks throughout their growth cycle and generate training data sets;

[0033] A prediction model building module, used to input the training data set into a deep learning model for training, and to build a growth prediction model for Muscovy ducks;

[0034] The prediction result analysis module is used to use the growth prediction model to input the weight and body size data of the Muscovy ducks in a specified period into the Muscovy duck growth prediction model for growth prediction, analyze the growth law of the Muscovy ducks, and obtain the Muscovy duck growth assessment report, which is used to screen the Muscovy ducks whose growth rate, weight and body size are not lower than the corresponding preset thresholds.

[0035] Furthermore, the steps of screening Muscovy ducks using the growth prediction model are as follows:

[0036] S41. Obtaining the weight and body size data of the Muscovy ducks in a specified period, cleaning and standardizing the data, and obtaining standard weight and body size data.

[0037] S42, inputting the standard body weight and body size data of step S11 into the growth prediction model, selecting the growth curve closest to the standard body weight and body size data for fitting, and obtaining a growth prediction curve.

[0038] S43, fitting and comparing the growth prediction curve of step S12 with the standard growth curve of breeding Muscovy ducks in the growth prediction model, and selecting Muscovy ducks whose growth rate, weight and body size are not lower than the standard growth curve.

[0039] Furthermore, the output content of the growth assessment report includes the weight growth curve, growth rate curve and growth cycle curve of the Muscovy duck.

[0040] The beneficial effects of the present invention are:

[0041] The data acquisition module of the present invention adopts a non-contact method for obtaining the body position image of the Muscovy duck, uses a depth camera to obtain the body position image of the Muscovy duck from multiple angles, uses a deep learning model to output the body size data of the Muscovy duck, and uses the data to establish a growth prediction model for screening the Muscovy ducks that meet the breeding requirements for key cultivation. The invention does not require manual intervention when obtaining the body size data of the Muscovy duck, which avoids the Muscovy ducks from having a large stress response in the process of obtaining data, and the use of a deep learning model can reduce the subjective error caused by manual measurement, and the automated measurement and screening reduces the cost of manual operation, which increases the development and growth of the scale of the Muscovy duck breeding industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0043] Figure 1 A top view of a measuring device for estimating the body size of a living Muscovy duck proposed by the present invention;

[0044] Figure 2 A first spatial side view of a measuring device for estimating the body size of a living Muscovy duck proposed by the present invention;

[0045] Figure 3 A second spatial side view of a measuring device for estimating the body size of a living Muscovy duck proposed by the present invention;

[0046] 1 box body, 2 circular column, 3 electric control door, 4 partition, 5 control main board, 6 depth camera, 7 distance sensor, 8 feeding trough, 9 conveyor track, 10 electric telescopic rod, 11 driving part, 12 voice recorder, 13 RFID electronic tag reading module, 14 weight sensor. DETAILED DESCRIPTION

[0047] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0048] The technical solution of the present invention is described in detail below in conjunction with the embodiments and drawings.

[0049] A system for estimating the body size of a Muscovy duck in vivo, the system comprising:

[0050] A data acquisition module, used to obtain body position images of Muscovy ducks;

[0051] like Figure 1 As shown, the data acquisition module includes a measuring device, which includes a box 1. A circular column 2, an electric control door 3, a partition 4 and a control main board 5 are provided in the box 1. The circular column 2 is arranged at the center of the box 1, and the control main board 5 is arranged on the circular column 2. The electric control door 3 is electrically connected to the control main board 5. The two sides of the electric control door 3 and the partition 4 are respectively connected to the circular column 2 and the wall of the box 1. The partition 4 divides the interior of the box 1 into multiple measuring units. The electric control door 3 divides the measuring unit into a first space and a second space. The first space and the second space are respectively provided with door-shaped holes opposite to the circular column 2 in the wall of the box 1.

[0052] like Figure 2 As shown, the first space includes a depth camera 6, a distance sensor 7, a feeding trough 8, a conveyor belt 9, a weight sensor 14 and an RFID electronic tag reading module 13. The conveyor belt 9 is electrically connected to the control mainboard 5. The depth camera 6, the distance sensor 7, the RFID electronic tag reading module 13 and the weight sensor 14 are respectively connected to the control mainboard 5 by signal. The depth camera 6 is respectively arranged at the top of the first space, the middle of the circular column 2 and the middle of the partition 4, for collecting body position images of the Muscovy duck at different angles. The distance sensor 7 is arranged next to each depth camera 6 for measuring the distance from the depth camera 6 to the Muscovy duck at different angles. The feeding trough 8 is arranged below the depth camera 6 in the middle of the circular column 2 and is detachably connected to the circular column 2 for attracting the Muscovy duck to enter the first space. The conveyor belt 9 is separately arranged at the bottom of each measuring unit for conveying the Muscovy duck from the first space to the second space. The weight sensor 14 and the RFID electronic tag reading module 13 are respectively arranged below the conveyor belt 9 in each first space.

[0053] In this embodiment, a circular box measuring device is provided to collect the body position images of the Muscovy duck at different angles, wherein the first space is provided with a feeding trough 8 to attract the Muscovy duck to enter the first space for image collection. The Muscovy duck wears an RFID electronic tag on its foot to store the unique identification information of the Muscovy duck. The feeding trough 8 attracts the Muscovy duck to enter the first space from the door-shaped hole. When the Muscovy duck steps on the conveying crawler 9, the RFID electronic tag reading module 13 reads the RFID electronic tag worn on the Muscovy duck's foot and sends the signal that the Muscovy duck enters the first space to the control mainboard 5. The control mainboard 5 controls the depth camera 6 to start collecting the body position images of the Muscovy duck, and controls the distance sensor 7 to start working. When the depth camera 6 takes each body position image, the distance sensor 7 assigns the distance from the Muscovy duck to the depth camera 6 to the body position image. The depth camera 6 and the weight sensor 14 send the acquired body position image, the weight data of the Muscovy duck, and the corresponding RFID electronic tag identity information to the control mainboard 5 for transmission to the next node.

[0054] The feeding trough 8 is set in the middle of the circular column 2. When the duck is feeding, its neck is naturally stretched, and the feeding trough 8 is fixedly placed, so that the duck is in a relatively static state, and the depth camera 6 at different angles is assisted to collect the body position images of the duck with good posture at different angles, which provides good data for the subsequent body size data generation and improves the system efficiency. After the depth camera 6 collects the body position image of the preset conditions, the control mainboard 5 controls the depth camera 6 to stop the image collection, and at the same time controls the electric control door 3 and the conveyor crawler 9 to open, and conveys the duck to the second space.

[0055] like Figure 3As shown, the second space includes a driving mechanism, which includes an electric telescopic rod 10, a driving part 11 and a mimeograph 12. The electric telescopic rod 10 is electrically connected to the control main board 5. The mimeograph 12 is arranged on the driving part 11 and is connected to the control main board 5 by signal. The electric telescopic rod 10 is arranged on the circular column 2 parallel to the ground. The driving part 11 is detachably connected to the electric telescopic rod 10, and is used to drive the Muscovy ducks out of the second space.

[0056] In this embodiment, when the RFID electronic tag reading module 13 and the weight sensor 14 cannot detect the RFID electronic tag and weight information, it means that the Muscovy duck is completely transferred to the second space by the conveyor track 9. The control main board 5 controls the electric control door 3 to close and the conveyor track 9 to stop, and at the same time controls the electric telescopic rod 10 to drive the driving part 11 to extend forward, which is used to drive the Muscovy duck out of the door-shaped hole in the second space. The driving part 11 is composed of mesh yarn wrapped around a hard coil. The driving part 11 composed of mesh yarn will not cause harm to the Muscovy duck when it touches the Muscovy duck, and can effectively drive the Muscovy duck out of the second space. When the electric telescopic rod 10 starts to work, the sound onomatopoeia 12 set on the driving part 11 simulates the call of the Muscovy duck, thereby avoiding the Muscovy duck from having greater emotional stress in the process of driving the Muscovy duck, so that the emotions of the Muscovy duck in the driving process are more stable.

[0057] In the above embodiments, the electric door, control main board, depth camera, distance sensor, conveyor belt, electric telescopic rod, voice recorder, RFID electronic tag reading module, RFID electronic tag and weight sensor can all adopt existing products on the market.

[0058] A wireless communication module, used to transmit the body position image acquired by the data acquisition module to the contour generation module, and transmit the weight data to the data analysis module;

[0059] The contour generation module is used to perform semantic segmentation based on the body image obtained by the data acquisition module using the A-Unet algorithm trained in the deep learning method to extract the edge contour curve of the Muscovy duck; compared with classic edge detection methods such as the Sobel operator, Canny operator, Laplace operator, etc., the method of segmenting the Muscovy duck contour curve and extracting the edge using the deep learning method does not require contour optimization and fitting, thereby improving the extraction accuracy and greatly shortening the contour extraction time.

[0060] The specific steps of using the trained A-Unet algorithm to perform semantic segmentation and extract the edge contour curve of the Muscovy duck are as follows:

[0061] S11, processing the original Muscovy duck body position image through the A-Unet model to generate a mask image, performing binarization processing on the mask image, and outputting a binarized image with mask information;

[0062] S12, using the binary image with mask information in step S11, and using the edge finding function findContours to extract the contour curve of the Muscovy duck;

[0063] S13. Use the drawContours function to draw the edge contour curve of the Muscovy duck.

[0064] The measuring point marking module is used to mark the body measurement points of the extracted Muscovy duck edge contour curve in OpenCV by using the dynamic grid method; in the image based on the Muscovy duck edge detection, the commonly used measuring point extraction methods include the manual marking method, the Harris corner point detection and the calculation of the maximum curvature by the curve fitting method, and the point located at the maximum curvature is the body measurement point. Considering the limitations of the aforementioned methods in detecting the measuring points of the Muscovy duck, the strategy based on the dynamic grid division of the present invention is used to effectively extract the body measurement points of the Muscovy duck.

[0065] The dynamic grid method is a grid-based image segmentation method that aims to extract the body measurement points of Muscovy ducks. This method divides the image into grid units by calculating the mean of the x-axis and y-axis coordinates of the pixel points of the Muscovy duck contour curve, and dynamically adjusts the grid units according to the feature information of the pixels in the unit until the stop condition is met. This strategy can more accurately depict the Muscovy duck contour curve, thereby extracting more accurate measurement points. In the processing of Muscovy duck contour images, the pixel point coordinate mean refers to the average value of the coordinate position of the contour line pixel points in the image. For Muscovy ducks of different body sizes, the extraction of each measurement point in the dynamic grid is based on the coordinates of the contour line pixel points. Therefore, for different Muscovy duck contour curves, the corresponding Muscovy duck body measurement points can be extracted.

[0066] Furthermore, the steps of marking the body measurement points using the dynamic grid method are as follows:

[0067] S21, using the bounding Rect function in OpenCV to obtain the maximum circumscribed matrix of the edge contour curve of the Muscovy duck from the binary edge contour curve of the Muscovy duck, wherein the maximum circumscribed matrix is ​​a horizontal rectangular rectangle including all key points in the image;

[0068] S22, dividing the image into grid units, and dynamically adjusting the grid units according to feature information of pixels in the units until the pixel coordinates of the edge contour curve of the Muscovy duck are extracted;

[0069] S23, calculating the mean of the x-axis and y-axis coordinates of the pixel points of the edge contour curve of the Muscovy duck, and extracting the corresponding body size measurement points of the Muscovy duck.

[0070] A data generation module is used to calculate and output the body size data of the Muscovy duck according to the ratio of the image pixels in the body image to the pixels of the actual length and the extracted body size measurement points;

[0071] The calibration of the depth camera 6 is an important step to achieve accurate measurement of the body size of the Muscovy duck through images. Its main purpose is to determine the internal parameters, external parameters and distortion coefficients of the depth camera 6 to achieve accurate measurement and accurate tracking of the object coordinates. The internal parameters of the depth camera 6 include inherent properties such as focal length, principal point position and distortion coefficient, among which the focal length is the basic parameter for determining imaging, and the principal point position is usually located at the center of the image. The external parameters of the depth camera 6 include its position and direction in the world coordinate system, which are used to map points in the world coordinate system to the depth camera 6 coordinate system, and finally to the image coordinate system. The distortion coefficient is the image distortion caused by the non-ideal characteristics of the depth camera 6 lens. By accurately determining the internal and external parameters of the depth camera 6, the distortion and error caused by the depth camera 6 can be effectively corrected, and the precision and accuracy of image measurement can be improved. The calibrated depth camera 6 can accurately measure the size of the object and calculate the actual size of the object.

[0072] The steps for calculating the body size are:

[0073] S31, calibrating the position of the depth camera 6, obtaining a pixel ratio of an image pixel to an actual length at each calibrated position, where the actual length is the distance from the Muscovy duck to the depth camera 6;

[0074] S32, calculating the pixel length between two body scale measurement points of the image using Euclidean distance calculation;

[0075] S33, converting the pixel length into the actual body size data of the Muscovy duck by using the pixel ratio calculated in step S31 and the pixel length calculated in step S32, and the calculation formula is:

[0076]

[0077] Where L is the body length, (X1, Y1)(X2, Y2) are the coordinates of the measurement points,

[0078] is the pixel length between two measuring points, and P is the pixel ratio.

[0079] The data analysis module is used to analyze the acquired body size data of Muscovy ducks, predict the growth trajectory of Muscovy ducks and screen breeding ducks based on the analysis results, and output a growth assessment report in the form of a chart.

[0080] A large amount of weight and body size data of Muscovy ducks are collected through the above modules, and then the RFID tags are used to track the weight and body size data of each Muscovy duck throughout its growth cycle, including the duckling stage, middle duck stage, reserve breeding duck stage, and adult duck stage; the collected data are feature labeled and enhanced to generate a training data set. The training data set is input into the deep learning model to learn the correlation model between the weight and body size of Muscovy ducks and their growth, that is, to build a growth prediction model for Muscovy ducks; the next time, the weight and body size data of the Muscovy ducks that need to be estimated are input into the growth prediction model. The model can predict the growth trajectory of the Muscovy ducks based on the weight and body size characteristics of the Muscovy ducks. If the Muscovy ducks meet the selection criteria for breeding ducks, they can be selected through the model. After the prediction results, the growth prediction model outputs a growth assessment report in the form of a chart. The growth trajectory curves of the Muscovy ducks in the report include weight growth curves, growth rate curves, and growth cycle curves. Through the report, the breeder can understand at which stage the weight of the Muscovy duck grows fastest, how long the rapid growth period lasts, and in which time period the standard for market release is reached. Based on the above information, breeders can select Muscovy ducks of similar stages for centralized breeding, and select Muscovy ducks with growth rate, weight and body size not lower than the corresponding preset thresholds for breeding. By distinguishing between breeding ducks and ordinary Muscovy ducks, breeders can adjust the ratio of feed for different Muscovy ducks in time, improve breeding efficiency and reduce breeding costs.

[0081] The data analysis module includes:

[0082] Data preprocessing module, used to track the weight and size data of Muscovy ducks throughout their growth cycle and generate training data sets;

[0083] A prediction model building module, used to input the training data set into a deep learning model for training, and to build a growth prediction model for Muscovy ducks;

[0084] The prediction result analysis module is used to use the growth prediction model to input the weight and body size data of the Muscovy ducks in a specified period into the Muscovy duck growth prediction model for growth prediction, analyze the growth law of the Muscovy ducks, and obtain the Muscovy duck growth assessment report, which is used to screen the Muscovy ducks whose growth rate, weight and body size are not lower than the corresponding preset thresholds.

[0085] By inputting the weight and body size of the duckling stage into the model, the growth prediction curve of the duck can be obtained; by tracking the weight and body size data of the same duck at different stages, the growth prediction model can also learn new data and continuously improve the prediction accuracy. When making predictions, the input data needs to be standardized to obtain standard weight and body size data to meet the data type that the model can recognize; the standard weight and body size data are input into the growth prediction model, and the model will find the closest growth curve to fit the standard weight and body size data according to the correlation between the weight and body size data of the duck, that is, to obtain the growth prediction curve of the duck. According to this principle, the more data the growth prediction model learns, the higher the accuracy will be. In the process of using the model, the prediction accuracy is also constantly improving. The growth prediction curve obtained by the growth prediction model will be further fitted and compared with the standard growth curve, and the standard growth curve will be adjusted according to actual needs; if the growth prediction curve fits the standard growth curve better, the probability that the duck can be selected as a breeder duck is greater, that is, the duck with a growth rate, weight and body size not lower than the standard growth curve will be selected by fitting and comparing the two curves.

[0086] The steps of screening Muscovy ducks using the growth prediction model are as follows:

[0087] S41. Obtaining the weight and body size data of the Muscovy ducks in a specified period, cleaning and standardizing the data, and obtaining standard weight and body size data.

[0088] S42, inputting the standard body weight and body size data of step S11 into the growth prediction model, selecting the growth curve closest to the standard body weight and body size data for fitting, and obtaining a growth prediction curve.

[0089] S43, fitting and comparing the growth prediction curve of step S12 with the standard growth curve of breeding Muscovy ducks in the growth prediction model, and selecting Muscovy ducks whose growth rate, weight and body size are not lower than the standard growth curve.

[0090] The output content of the growth assessment report includes the weight growth curve, growth rate curve and growth cycle curve of the Muscovy duck.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A system for estimating the body size of a Muscovy duck in vivo, characterized in that: The system comprises: A data acquisition module, used to obtain body position images of Muscovy ducks; A wireless communication module, used to transmit the body position image acquired by the data acquisition module to the contour generation module, and transmit the weight data to the data analysis module; The contour generation module is used to perform semantic segmentation based on the body position image obtained by the data acquisition module using the A-Unet algorithm trained in deep learning to extract the edge contour curve of the Muscovy duck; The measurement point marking module is used to mark the body measurement points of the extracted edge contour curve of the Muscovy duck using the dynamic grid method in OpenCV; A data generation module is used to calculate and output the body size data of the Muscovy duck according to the ratio of the image pixels in the body image to the pixels of the actual length and the extracted body size measurement points; The data analysis module is used to analyze the acquired body size and weight data of Muscovy ducks, predict the growth trajectory of Muscovy ducks and screen breeding ducks based on the analysis results, and output a growth assessment report in the form of a chart.

2. A system for estimating the body size of a Muscovy duck in vivo according to claim 1, characterized in that: The data acquisition module includes a measuring device, which includes a box body. A circular column, an electric control door, a partition and a control main board are arranged in the box body. The circular column is arranged in the center of the box body, and the control main board is arranged on the circular column. The electric control door is electrically connected to the control main board. The two sides of the electric control door and the partition are respectively connected to the circular column and the box body wall. The partition divides the interior of the box body into multiple measuring units, and the electric control door divides the measuring unit into a first space and a second space. The first space and the second space are respectively provided with door-shaped holes opposite to the circular column in the box body wall.

3. A system for estimating the body size of a Muscovy duck in vivo according to claim 2, characterized in that: The first space includes a depth camera, a distance sensor, a feeding trough, a conveyor belt, a weight sensor and an RFID electronic tag reading module. The conveyor belt is electrically connected to the control mainboard. The depth camera, the distance sensor, the RFID electronic tag reading module and the weight sensor are respectively connected to the control mainboard signal. The depth cameras are respectively arranged at the top of the first space, the middle of the circular column and the middle of the partition, and are used to collect body position images of the Muscovy duck at different angles. The distance sensor is arranged next to each depth camera, and is used to measure the distance from the depth camera to the Muscovy duck at different angles. The feeding trough is arranged below the depth camera in the middle of the circular column and is detachably connected to the circular column to attract the Muscovy duck to enter the first space. The conveyor belt is separately arranged at the bottom of each measuring unit, and is used to convey the Muscovy duck from the first space to the second space. The weight sensor and the RFID electronic tag reading module are respectively arranged below the conveyor belt of each first space.

4. The system for estimating the body size of a Muscovy duck in vivo according to claim 2, characterized in that: The second space includes a driving mechanism, which includes an electric telescopic rod, a driving part and a mimeograph, the electric telescopic rod is electrically connected to the control mainboard, the mimeograph is arranged on the driving part and is connected to the control mainboard signal, the electric telescopic rod is arranged on a circular column parallel to the ground, and the driving part is detachably connected to the electric telescopic rod, and is used to drive the Muscovy ducks out of the second space.

5. The system for estimating the body size of a Muscovy duck in vivo according to claim 1, characterized in that: The specific steps of using the trained A-Unet algorithm to perform semantic segmentation and extract the edge contour curve of the Muscovy duck are as follows: S11, processing the original Muscovy duck body position image through the A-Unet model to generate a mask image, performing binarization processing on the mask image, and outputting a binarized image with mask information; S12, using the binary image with mask information in step S11, and using the edge finding function findContours to extract the contour curve of the Muscovy duck; S13. Use the drawContours function to draw the edge contour curve of the Muscovy duck.

6. A system for estimating the body size of a Muscovy duck in vivo according to claim 5, characterized in that: The steps of marking body measurement points using the dynamic grid method are as follows: S21, using the bounding Rect function in OpenCV to obtain the maximum circumscribed matrix of the edge contour curve of the Muscovy duck from the binary edge contour curve of the Muscovy duck, wherein the maximum circumscribed matrix is ​​a horizontal rectangular rectangle including all key points in the image; S22, dividing the image into grid units, and dynamically adjusting the grid units according to feature information of pixels in the units until the pixel coordinates of the edge contour curve of the Muscovy duck are extracted; S23, calculating the mean of the x-axis and y-axis coordinates of the pixel points of the edge contour curve of the Muscovy duck, and extracting the corresponding body size measurement points of the Muscovy duck.

7. The system for estimating the body size of a Muscovy duck in vivo according to claim 3, characterized in that: The steps for calculating the body size are: S31, calibrating the position of the depth camera, and obtaining a pixel ratio of an image pixel to an actual length at each calibrated position, where the actual length is the distance from the Muscovy duck to the depth camera; S32, calculating the pixel length between two body scale measurement points of the image using Euclidean distance calculation; S33, converting the pixel length into the actual body size data of the Muscovy duck by using the pixel ratio calculated in step S31 and the pixel length calculated in step S32, and the calculation formula is: Where L is the body length, (X1, Y1)(X2, Y2) are the coordinates of the measurement points, is the pixel length between two measuring points, and P is the pixel ratio.

8. The system for estimating the body size of a Muscovy duck in vivo according to claim 1, characterized in that: The data analysis module includes: Data preprocessing module, used to track the weight and size data of Muscovy ducks throughout their growth cycle and generate training data sets; A prediction model building module, used to input the training data set into a deep learning model for training, and to build a growth prediction model for Muscovy ducks; The prediction result analysis module is used to use the growth prediction model to input the weight and body size data of the Muscovy ducks in a specified period into the Muscovy duck growth prediction model for growth prediction, analyze the growth law of the Muscovy ducks, and obtain the Muscovy duck growth assessment report, which is used to screen the Muscovy ducks whose growth rate, weight and body size are not lower than the corresponding preset thresholds.

9. A system for estimating the body size of a Muscovy duck in vivo according to claim 8, characterized in that: The steps of screening Muscovy ducks using the growth prediction model are as follows: S41. Obtaining the weight and body size data of the Muscovy ducks in a specified period, cleaning and standardizing the data, and obtaining standard weight and body size data. S42, inputting the standard body weight and body size data of step S11 into the growth prediction model, selecting the growth curve closest to the standard body weight and body size data for fitting, and obtaining a growth prediction curve. S43, fitting and comparing the growth prediction curve of step S12 with the standard growth curve of breeding Muscovy ducks in the growth prediction model, and selecting Muscovy ducks whose growth rate, weight and body size are not lower than the standard growth curve.

10. The system for estimating the body size of a Muscovy duck in vivo according to claim 8, characterized in that: The output content of the growth assessment report includes the weight growth curve, growth rate curve and growth cycle curve of the Muscovy duck.

Images