WO2020078111A1

WO2020078111A1 - Weight measurement method and device, and computer readable storage medium

Info

Publication number: WO2020078111A1
Application number: PCT/CN2019/103274
Authority: WO
Inventors: 王博; 李春华
Original assignee: 京东数字科技控股有限公司
Priority date: 2018-10-17
Filing date: 2019-08-29
Publication date: 2020-04-23
Also published as: CN109459119A; CN109459119B

Abstract

A weight measurement method and device (4), and a computer readable storage medium. The weight measurement method comprises: obtaining image information to be monitored for an object to be measured, and carrying out image recognition on said image information to obtain a contour of said object (101); determining a posture of said object on the basis of the contour of said object (102); and calculating the weight of said object on the basis of the posture and a weight mapping matrix of said object (103). The weight mapping matrix is obtained by means of pre-training.

Description

Weight measuring method, equipment and computer readable storage medium

Cross-reference of related applications

This application is based on a Chinese patent application with an application number of 201811210433.4 and an application date of October 17, 2018, and claims the priority of the Chinese patent application. The entire content of the Chinese patent application is hereby incorporated into this application by way of introduction.

Technical field

The present invention relates to image processing technology in the field of communications, and in particular, to a weight measurement method, device, and computer-readable storage medium.

Background technique

China has become the largest country in pork production, with pork production and consumption accounting for more than half of the world. Based on this, with the continuous development of modern animal husbandry, some companies have begun to apply automation control, computer vision and other technologies in the field of pig farming, in order to improve the management and control of the farm, improve the production environment of the farm, and increase the efficiency of the staff . During the breeding process, regular changes in the weight of the pig, and then adjustment of the pen position, feeding amount, etc., are of great significance for improving the breeding efficiency.

The existing pig body weight measurement methods are mainly divided into contact and non-contact types: contact measurement methods are further divided into direct measurement methods and indirect measurement methods; direct measurement mainly uses tools such as scales and electronic scales; indirect measurement mainly through the Pig body length, chest circumference, hip circumference and other body size indicators are measured, and then the weight of the pig is obtained through empirical formulas; the non-contact measurement method mainly relies on the camera to collect the image of the pig, and then uses the digital image processing technology to Body weight is estimated.

However, the inventor found that the traditional contact method in the prior art has a time-consuming and labor-intensive measurement method, and there are large errors; at the same time, the inventor found that the non-contact method in the prior art has at least a limited use range and needs to be based on Only human-computer interaction can be realized, which leads to greater maintenance costs and operational difficulties.

Summary of the invention

In order to solve the above technical problems, the embodiments of the present invention are expected to provide a weight measurement method, device, and computer-readable storage medium, which solves the problem of large measurement error and high cost in measuring pig weight in the prior art. The problem has realized the accurate measurement of the weight of the pig, and reduced the operation difficulty and maintenance cost; at the same time, it has universal applicability.

The technical solution of the present invention is implemented as follows:

A method for measuring body weight, the method includes: acquiring image information to be monitored for an object to be measured, and performing image recognition on the image information to be monitored to obtain an outline of the object to be measured; based on the outline of the object to be measured Determine the posture of the object to be measured; calculate the weight of the object to be measured based on the posture of the object to be measured and the weight mapping matrix; wherein the weight mapping matrix is obtained by pre-training.

A weight measurement device, the device includes: a processor, a memory, and a communication bus;

The communication bus is used to implement a communication connection between the processor and the memory;

The processor is used to execute a weight measurement program stored in the memory to achieve the following steps:

Acquiring image information to be monitored for the object to be measured, and performing image recognition on the image information to be monitored to obtain the contour of the object to be measured;

Determine the posture of the object to be measured based on the outline of the object to be measured;

Calculate the weight of the object to be measured based on the posture and weight mapping matrix of the object to be measured; wherein the weight mapping matrix is obtained by pre-training.

A computer-readable storage medium stores one or more programs, and the one or more programs may be executed by one or more processors to implement the steps of the weight measurement method described above.

Because the image information to be monitored of the object to be measured is acquired, and the image of the image to be monitored is image-recognized to obtain the contour of the object to be measured, then the posture of the object to be measured is determined based on the contour of the object to be measured, and finally based on the posture and The weight mapping matrix calculates the weight of the object to be measured. In this way, without the need for human participation, it is only necessary to extract the contour of the object to be measured from the image information of the object to be measured, and based on the contour of the object to be measured and pre-training The obtained weight mapping matrix can realize the measurement of the weight of the object to be measured, which solves the problems of large measurement error and high cost when measuring the weight of the pig in the prior art, and realizes the accuracy of the weight of the pig Measurement, and reduces the difficulty of operation and maintenance costs; at the same time, it has universal applicability.

BRIEF DESCRIPTION

FIG. 1 is a schematic flowchart of a weight measurement method according to an embodiment of the present invention;

2 is a schematic flowchart of another weight measurement method provided by an embodiment of the present invention;

3 is a schematic diagram of a standard posture of an object to be measured provided by an embodiment of the present invention;

FIG. 4 is a schematic flowchart of another weight measurement method according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a weight measurement device according to an embodiment of the present invention.

detailed description

The technical solutions in the embodiments of the present invention will be described clearly and completely in conjunction with the drawings in the embodiments of the present invention.

An embodiment of the present invention provides a weight measurement method. Referring to FIG. 1, the method includes the following steps:

Step 101: Acquire image information to be monitored for the object to be measured, and perform image recognition on the image information to be monitored to obtain the contour of the object to be measured.

Among them, step 101 acquires the image information to be monitored for the object to be measured, and performs image recognition on the image information to be monitored to obtain the contour of the object to be measured can be realized by the weight measuring device; Obtained; the image information to be monitored may include one image information or multiple image information, which can be determined according to actual application scenarios. The contour of the object to be measured may be obtained by performing image recognition on the object to be measured included in the image information to be monitored.

Step 102: Determine the posture of the object to be measured based on the contour of the object to be measured.

Wherein, in step 102, determining the posture of the object to be measured based on the contour of the object to be measured can be implemented by the weight measuring device; matching the image corresponding to the contour of the object to be measured with the image of the standard posture, and determining according to the result of the matching process The posture of the object to be measured.

Step 103: Based on the posture and weight mapping matrix of the object to be measured, calculate the weight of the object to be measured.

Among them, the weight mapping matrix is obtained by pre-training.

It should be noted that in step 103, based on the posture of the object to be measured and the weight mapping matrix, calculating the weight of the object to be measured can be achieved by the weight measuring device; the weight measuring device can obtain the correspondence with the posture of the object to be measured according to the posture of the object to be measured Weight mapping matrix, and calculate the weight of the object to be measured according to the mask matrix and the weight mapping matrix of the object to be measured.

The weight measurement method provided by the embodiment of the present invention obtains the image information to be monitored, and performs image recognition on the image information to obtain the contour of the object to be measured, and then determines the posture of the object to be measured based on the contour of the object to be measured, and finally The measurement object's posture and weight mapping matrix calculate the weight of the object to be measured. In this way, without the need for human participation, only the contour of the object to be measured needs to be extracted from the image information of the object to be measured, and based on the object to be measured The contour and the weight mapping matrix obtained by pre-training can realize the measurement of the weight of the object to be measured, which solves the problems of large measurement error and high cost in the prior art when measuring the weight of the pig, and realizes the Accurate measurement of pig body weight reduces operation difficulty and maintenance costs; at the same time, it has universal applicability.

Based on the foregoing embodiments, the embodiments of the present invention provide a weight measurement method. Referring to FIG. 2, the method includes the following steps:

Step 201: The weight measurement device acquires image information to be monitored for the object to be measured through the image collector.

Among them, the image information to be monitored may be collected and realized through an image collector; the image collector may be a monocular camera, such as a monocular camera. In a feasible implementation manner, the image information to be monitored may include a plurality of images to be monitored; of course, the image information to be monitored may also refer to the video to be monitored.

Step 202: The weight measuring device acquires a preset number of objects to be trained that meet a predetermined condition.

Among them, the posture of the object to be trained is a standard posture.

It should be noted that the preset number is an amount that can be adjusted according to actual application requirements and application scenarios, and the preset number is related to the number of standard gestures acquired; in a feasible implementation manner, the preset The number can be an integer multiple of the number of standard postures obtained; if the number of standard postures obtained is two (that is, including two standard postures), the preset number can be an integer multiple of 2; for example, the preset The number can be 1000, where the objects to be trained are all those with a standard posture; for example, the objects to be trained can include 500 to be trained as the first standard posture, and can also include 500 to be the second posture Subjects to be trained in standard poses.

In addition, the predetermined condition may mean that the object to be trained covers various parts in the collected picture, and the object to be trained covers various types of objects. That is to say, the object to be trained can be obtained by monitoring the object that needs to be trained in advance through the camera, and then selecting a preset number of objects from a large number of objects and covering objects in various parts of the picture collected by the camera .

Step 203: The weight measuring device obtains the number of objects to be trained included in each pixel in the image information to be monitored, and generates a first target matrix based on the number of objects to be trained.

The number of objects to be trained included on each pixel in the image information to be monitored may refer to the number of objects to be trained that have appeared on each pixel in the pixels included in the picture of the image to be monitored.

The first target matrix may be generated based on the acquired number of objects to be trained and the first matrix set initially, and the number of rows and columns of the first matrix is the same as the value indicated by the size of the image information to be monitored. In a feasible implementation manner, the first target matrix may be used to record the number of objects to be trained that appear at each pixel.

Step 204: The weight measurement device generates a second target matrix based on the weight of the object to be trained and the pixel area of the covered pixels.

Among them, the weight of the object to be trained is obtained after the weight measuring device measures the object to be trained in real time; meanwhile, the pixel area is the total pixel area of the pixel points occupied by the object to be trained in the image information to be monitored by the weight measuring device owned.

The second target matrix may be generated based on the acquired weight, pixel area and initial setting of the second matrix of the object to be trained. The number of rows and columns of the second matrix is the same as the value indicated by the size of the image information to be monitored. In a feasible implementation manner, the second target matrix may be used to record the cumulative weight weight of the object to be trained that has appeared at each pixel.

Step 205: The weight measuring device generates a weight mapping matrix based on the second target matrix and the first target matrix.

After obtaining the first target matrix and the second target matrix, the second target matrix can be divided by the first target matrix, and the obtained quotient is the weight mapping matrix.

Step 206: The weight measuring device uses a specific image segmentation algorithm to perform image recognition on the image information to be monitored, and determines the contour of the object to be measured from the image information to be monitored.

Among them, the specific image segmentation algorithm can refer to Masked Region Convolutional Neural Network (Mask Regions With Convolutional Neural Network, Mask R-CNN). The contour of the object to be measured can be obtained by segmenting and edge detecting the object in the image information to be monitored using the Mask R-CNN algorithm.

Step 207: The weight measuring device acquires a standard posture for the object to be measured.

Among them, the standard posture can be set according to the actual application requirements and application scenarios, combined with the physical object of the measurement object; the standard posture can be a posture that each object to be measured has and is generally applicable, in the embodiment of the present invention The standard posture obtained does not limit the specific number; if the object to be measured is a pig that needs to be measured, the standard posture obtained in this embodiment of the present invention may include two types (that is, two); as shown in FIG. 3, the standard posture Specifically, it may include two postures: the first standard posture A shown in FIG. 3a of FIG. 3 and the second standard posture B shown in FIG. 3b of FIG. 3; where the first standard posture may be Standing posture, the second standard posture can be a lateral lying posture.

Step 208: The weight measuring device determines the posture of the object to be measured based on the standard posture and the contour of the object to be measured.

Among them, the area of the same part of the image corresponding to the contour of the object to be measured and the image corresponding to the standard posture is obtained according to the contour of the object to be measured, and the matching degree between the two images mentioned above is calculated according to the area, and then according to the matching The relationship between the degree and the preset threshold determines the posture of the object to be measured.

Step 209: The weight measuring device calculates the weight of the object to be measured based on the posture of the object to be measured and the weight mapping matrix.

Among them, the object to be trained and the object to be measured in the embodiment of the present invention may be any object that needs to be calculated for weight; in a feasible implementation manner, the object to be trained and the object to be measured may include different types of objects that need to be calculated for weight The animal may be, for example, poultry.

It should be noted that the description of the same steps and the same content in this embodiment and other embodiments can refer to the descriptions in other embodiments, which will not be repeated here.

The weight measurement method provided by the embodiment of the present invention only needs to extract the contour of the object to be measured from the image information of the object to be measured for the object to be measured without manual participation, and based on the contour of the object to be measured and pre-training The obtained weight mapping matrix can realize the measurement of the weight of the object to be measured. At the same time, because the image to be monitored is collected using a monocular camera, it solves the large measurement error existing in the prior art when measuring the weight of the pig. The problem of higher cost has realized the accurate measurement of the pig's weight, and reduced the operation difficulty and maintenance cost; at the same time, it has universal applicability.

Based on the foregoing embodiment, an embodiment of the present invention provides a weight measurement method. Referring to FIG. 4, the method includes the following steps:

Step 301: The weight measuring device obtains image information to be monitored for the object to be measured through the image collector.

Step 302: The weight measurement device acquires a preset number of objects to be trained that meet a predetermined condition.

Step 303: The weight measuring device obtains the number of objects to be trained included in each pixel in the image information to be monitored, and generates a first target matrix based on the number of objects to be trained.

Among them, step 303 can be implemented in the following manner:

Step 303a: The weight measuring device sets the first matrix of M * N.

Among them, M * N is the size of the image information to be monitored.

In other embodiments of the present invention, the value of M may be the length value corresponding to the size of the image information to be monitored, and the value of N may be the width value corresponding to the size of the image information to be monitored. In a feasible implementation manner, if the size of the image information to be monitored is 1280 * 1024, the value of M may be, and the value of N may be 1024, that is, the first matrix is a matrix of 1280 * 1024. It should be noted that the first matrix is an initialized matrix, and the values of corresponding elements are all 0.

Step 303b: The weight measuring device sequentially traverses the outline of the preset number of objects to be trained, and processes the outline of each object to be trained to obtain the target part of each object to be trained.

Among them, the contour of the object to be trained may be obtained by segmenting the acquired object to be trained and edge detection using the Mask R-CNN algorithm.

The image information to be monitored in the embodiment of the present invention may be a two-dimensional planar image, and the object to be trained may be a pig body. Considering that in the two-dimensional planar image, the main body weight of the pig body is distributed in the pig body, legs, tail and head The weight distribution is relatively small. In order to improve the accuracy of the algorithm, the image processing method can be used to remove the pig body part with a more prominent pig body contour; further, only the pig body part area is used as the training weight mapping matrix and the effective area for estimating weight. That is to say, the target portion may be determined according to the actual object of the object specifically referred to by the object to be trained, and may be an effective portion that can represent the weight of the object to be trained. In a feasible implementation manner, if the object to be trained can be a pig body, then the target part can be a pig body part.

Step 303c: The weight measuring device obtains the quantity value of the target portion of the object to be trained included on each pixel in the image information to be monitored.

Wherein, the quantity value of the target portion of the object to be trained included on each pixel in the image information to be monitored in the embodiment of the present invention may refer to each pixel in the pixel included in the picture of the image information to be monitored The number of target parts of the object to be trained has appeared on the point.

Step 303d. The weight measuring device assigns the quantity value to the first matrix according to the correspondence between the pixel points and the first matrix to obtain the first target matrix.

Among them, there is a correspondence between pixels and elements in the first matrix.

In other embodiments of the present invention, because the specifications of the first matrix are the same as the size of the image information to be monitored, there is a one-to-one correspondence between each pixel in the image information to be monitored and the elements in the first matrix . The first target matrix may be an M * N matrix obtained by updating the values of the elements in the first matrix using quantity values. In a feasible implementation manner, if the number of target parts of the object to be trained that has appeared on a pixel in the image information to be monitored is 400, then the corresponding element in the first matrix corresponding to this pixel The value is 400.

It should be noted that if the first matrix is W _n (x, y), the target part of the object to be trained is P ′ _i ; then, the first target matrix can be generated using the following formula (1):

Among them, x and y respectively represent the horizontal and vertical coordinates of the pixel; formula (1) indicates that if the target part of the object to be trained appears once on the pixel, the value of the corresponding element in the first matrix corresponding to the pixel Add 1; if the pixel is the target part where the object to be trained appears, the value of the corresponding element in the first matrix corresponding to the pixel remains unchanged; the updated matrix after the assignment of all elements of the first matrix is The first target matrix.

Step 304: The weight measuring device generates a second target matrix based on the weight of the object to be trained and the pixel area of the covered pixels.

Step 304a: The weight measuring device sets the second matrix of M * N.

The specifications of the second matrix are the same as the specifications of the first matrix, and the second matrix is also a matrix formed after initialization, and the values of the corresponding elements are all 0.

Step 304b: The weight measuring device measures the weight of each object to be trained, and calculates the pixel area of the pixels covered by the target part of each object to be trained in the image information to be monitored.

Among them, the weight corresponding to the training can be obtained by weighing the actual weight of 500 pigs with the first standard posture (or the second standard posture) used for training using a scale, electronic scale or other tools . The pixel area may refer to the area of the area formed by all pixels covered by the pig body part of each pig in the image information to be monitored.

Step 304c: The weight measuring device calculates the weight value corresponding to each pixel in the pixels covered by each object to be trained based on the weight and pixel area of each object to be trained.

The weight weight corresponding to each pixel in the pixels covered by each object to be trained may be obtained by dividing the actual weight of each object to be trained by the pixel area.

Step 304d: The weight measuring device adds the weight weights corresponding to the same pixels among the pixels covered by the preset number of objects to be trained to obtain the weight weight corresponding to each pixel in the image information to be monitored.

In a feasible implementation manner, if there are 400 objects to be trained on a pixel in the image information to be monitored, the weight weights of the 400 objects to be trained are a1, a2, a3 ... a400, then The weight value of the corresponding pixel in the image information to be monitored is a1 + a2 + a3 + ... + a400.

Step 304e: The weight measurement device assigns weight weights to the second matrix according to the correspondence between pixels and the second matrix to obtain a second target matrix.

Since the specifications of the second matrix are the same as the size of the image information to be monitored, each pixel in the image information to be monitored has a one-to-one correspondence with the elements in the second matrix. The second target matrix may be an M * N matrix obtained by updating the values of elements in the second matrix using weight weights. In a feasible implementation manner, if the weight value corresponding to a certain pixel in the image information to be monitored is 300, then the value of the element corresponding to this pixel in the corresponding first matrix is 300.

It should be noted that if the second matrix is W _m (x, y), the actual weight of the object to be trained is M _i , and the pixel area corresponding to the target portion of the object to be trained is S _i ; then, the second target matrix can be used The following formula (2) is generated:

Among them, formula (2) indicates that the value of the corresponding element in the second matrix corresponding to each pixel covered by the target part of the training object is added to the quotient obtained by dividing the actual weight of the training object by the pixel area; If the pixel is not covered by the target part of the training object, the value of the corresponding element in the second matrix is unchanged; the updated matrix after the assignment of all elements of the second matrix is the second target matrix.

Step 305: The weight measuring device generates a weight mapping matrix based on the second target matrix and the first target matrix.

It should be noted that, in the process of generating the weight mapping matrix, the objects to be trained are classified according to the pre-selected standard posture; if the standard posture includes two types, the objects to be trained need to be divided into two types at this time; the first type is the posture as The first type of standard posture to be trained, the second type is the type of posture to be trained. In the subsequent generation of the weight mapping matrix, the first weight mapping matrix is trained according to the first type of object to be trained, and the second weight mapping matrix is obtained according to the second type of object to be trained. Of course, in practical applications, only one weight mapping matrix or more weight mapping matrices can be obtained, which can be determined according to the number of standard postures acquired.

Step 306: The weight measuring device uses a specific image segmentation algorithm to perform image recognition on the image information to be monitored, and determines the contour of the object to be measured from the image information to be monitored.

Step 307: The weight measuring device acquires a standard posture for the object to be measured.

Step 308: The weight measuring device acquires an image with a standard posture, and converts the format of the image with a standard posture into a preset format to obtain a first image.

Among them, the preset format is a binarization format; binarization refers to the image obtained by setting the background part of the image to 1 and the foreground part to 0; the first image may be an image whose pose is extracted as a standard pose After binarization, and then size scaling is obtained.

Step 309: The weight measuring device acquires an image corresponding to the contour of the object to be measured, and converts the format of the image corresponding to the contour of the object to be measured into a preset format to obtain a second image.

The second image may be obtained by binarizing the image corresponding to the contour of the object to be measured and then scaling the size; it should be noted that if the image information to be monitored includes multiple objects to be measured, then this time You can get multiple second images corresponding to the number.

It should be noted that the size scaling ratio when acquiring the first image and the second image may be the same, for example, the size scaling ratio of the image with the first standard posture may be 200 * 80, and the size scaling of the image with the second standard posture The ratio can be 200 * 100.

Step 310: The weight measuring device calculates the degree of matching between the second image and the first image.

Wherein, the calculation of the matching degree between the second image and the first image in step 310 may be implemented in the following manner:

Step 310a. The weight measuring device acquires a first area corresponding to the same area in the first image and the second image.

The first area may be obtained by comparing the first image with the second image and calculating the area of the area corresponding to the same part in the first image and the second image.

Step 310b. The weight measuring device calculates the sum of the area of the first image and the area of the second image to obtain the second area.

The second area may be obtained by adding the area of the first image and the area of the second image.

Step 310c. The weight measuring device calculates the ratio of the first area to the second area to obtain the matching degree of the second image and the first image.

Where, if the first image identifier is M _ref and the second image identifier is m, then the matching degree R between the second image and the first image can be calculated using formula (3):

Step 311: If the degree of matching between the second image and the first image is greater than or equal to a preset threshold, the weight measuring device determines that the posture of the object to be measured is a standard posture corresponding to the first image with a degree of matching greater than or equal to the preset threshold.

Where, if the standard posture includes the first standard posture and the second standard posture, the first image includes two images; if the second image matches the image of the first standard posture in the first image, the matching degree is greater than or Equal to the preset threshold, then the posture of the object to be measured is the first standard posture; if the degree of matching between the second image and the image of the second standard posture in the first image is greater than or equal to the preset threshold, then the object to be measured The posture is the second standard posture; the preset threshold can be set according to the actual application scenario and specific requirements.

Step 312: If the matching degree is less than the preset threshold, the weight measuring device rotates the second image according to the predetermined direction and the preset angle to obtain the third image, and calculates the matching degree of the third image and the first image.

Wherein, the predetermined direction and the preset angle may be a preset direction and angle; in a feasible implementation manner, the predetermined direction may be clockwise or counterclockwise, and the preset angle may be 45 °.

Step 313: If the matching degree between the third image and the first image is less than a preset threshold, the weight measuring device rotates the third image according to a predetermined direction and a predetermined angle.

Wherein, if the matching degree between the third image and the first image is greater than or equal to a preset threshold, the weight measuring device determines that the posture of the object to be measured is a standard posture corresponding to the first image with a matching degree greater than or equal to the preset threshold.

If the match between the second image obtained initially and the two images in the first image is less than the preset threshold, then rotate the second image clockwise (or counterclockwise) by 45 ° to obtain the third image, if the third image The degree of matching with the image in the first image whose pose is the first standard posture is greater than or equal to the preset threshold, then the posture of the object to be measured is the first standard posture; if the posture in the second image and the first image is the second The matching degree of the image of the standard posture is greater than or equal to the preset threshold, then the posture of the object to be measured is the second standard posture.

Step 314: If after one rotation according to a predetermined direction and a predetermined angle, the degree of matching between each rotated image and the first image is less than a preset threshold, the weight measuring device determines that the posture of the object to be measured is an invalid posture.

Among them, if the match between the third image and the two images in the first image is less than the preset threshold, then continue to rotate the second image clockwise (or counterclockwise) by 45 ° to obtain a new image, if the new The matching degree between the image and any one of the first images is greater than or equal to the preset threshold, at this time it can be obtained that the posture of the object to be measured is the standard posture corresponding to the image with the matching degree greater than or equal to the preset threshold; The matching degree of any image in the first image is less than the preset threshold, then continue to rotate the second image in a clockwise (or counterclockwise) direction by 45 °; if the second image is rotated once (360 °), rotate The matching degree between all the images obtained in the process and the two images in the first image is less than the preset threshold. At this time, the posture of the object to be measured is determined to be an invalid posture, and the corresponding contour is marked as an invalid contour. Subject runs subsequent weight estimates.

Step 315: The weight measuring device calculates the weight of the object to be measured based on the posture of the object to be measured and the weight mapping matrix.

Step 315a. If the posture of the object to be measured is a standard posture, obtain the position of the pixel point covered by the target part of the object to be measured in the image information to be monitored.

Step 315b: Generate an M * N mask matrix based on the positions of the pixels.

Among them, the mask matrix may be generated in the following manner: first initialize a third matrix of M * N, the specifications of the third matrix are the same as the specifications of the first matrix, and are formed after initialization, and their corresponding The values of the elements are all 0; calculate the position of the pixel covered by the target part of the object to be measured in the image information to be monitored, and mark the value of the element corresponding to the pixel in the third matrix as 1, which is the same as the object to be measured The value of the element corresponding to the pixel that is not covered by the target part is recorded as 0, and finally the M * N mask matrix is generated. Wherein, each pixel in the image information to be monitored has a one-to-one correspondence with the elements in the third matrix.

Step 315c: Based on the mask matrix and the weight mapping matrix, calculate the weight of the object to be measured.

Wherein, the mask matrix is multiplied by the corresponding weight mapping matrix, and then the value of each element in the obtained new matrix is added, and the sum is the weight of the object to be measured. In other embodiments of the present invention, if the posture of the object to be measured is the first standard posture, the first weight mapping matrix in the weight mapping matrix may be multiplied by the mask matrix, and then the new matrix obtained The value of each element in is added to obtain the weight of the object to be measured; if the posture of the object to be measured is the second standard posture, then the second weight mapping matrix in the weight mapping matrix can be multiplied by the mask matrix , And then add the value of each element in the obtained new matrix to obtain the weight of the object to be measured.

If multiple objects to be measured are identified from the image information to be monitored, the weight measurement method provided in the embodiment of the present invention may be used to measure the weight of each object to be measured.

The weight measurement method provided by the embodiment of the present invention only needs to extract the contour of the object to be measured from the image information of the object to be measured for the object to be measured without manual participation, and based on the contour of the object to be measured and pre-training The obtained weight mapping matrix can realize the measurement of the weight of the object to be measured, which solves the problems of large measurement error and high cost when measuring the weight of the pig in the prior art, and realizes the accuracy of the weight of the pig Measurement, and reduces the difficulty of operation and maintenance costs; at the same time, it has universal applicability.

Based on the foregoing embodiments, embodiments of the present invention provide a weight measurement device, which can be applied to a weight measurement method provided in the embodiments corresponding to FIGS. 1-2 and 4, as shown in FIG. 5, The weight measuring device 4 may include: a processor 41, a memory 42, and a communication bus 43, wherein:

The communication bus 43 is used to realize the communication connection between the processor 41 and the memory 42;

The processor 41 is used to execute the weight measurement program stored in the memory 42 to achieve the following steps: acquiring image information to be monitored for the object to be measured, and performing image recognition on the image information to obtain the outline of the object to be measured; based on the object to be measured The contour of determines the posture of the object to be measured; based on the posture of the object to be measured and the weight mapping matrix, calculates the weight of the object to be measured; wherein the weight mapping matrix is obtained by pre-training.

In other embodiments of the present invention, the processor 41 is used to execute the acquisition of the image information to be monitored stored in the memory 42 and perform image recognition on the image information to be monitored to obtain the contour of the object to be measured, in order to achieve the following steps: through the image collector Obtain the image information to be monitored; use a specific image segmentation algorithm to perform image recognition on the image information to be monitored, and determine the contour of the object to be measured from the image information to be monitored.

In other embodiments of the present invention, the processor 41 is used to execute the determination of the posture of the object to be measured based on the contour of the object to be measured stored in the memory 42 to achieve the following steps: acquiring a standard posture for the object to be measured; based on the standard posture And the contour of the object to be measured, determine the posture of the object to be measured.

In other embodiments of the present invention, the processor 41 is used to execute the standard posture and the contour of the object to be measured stored in the memory 42 to determine the posture of the object to be measured, so as to achieve the following steps: acquiring an image with a standard posture, Convert the format of the image with the standard pose to the preset format to obtain the first image; obtain the image corresponding to the contour of the object to be measured, and convert the format of the image corresponding to the contour of the object to be measured to the preset format to obtain the second Image; calculate the degree of matching between the second image and the first image; if the degree of matching between the second image and the first image is greater than or equal to the preset threshold, determine the posture of the object to be measured is the first degree of matching greater than or equal to the preset threshold The standard pose corresponding to the image.

In other embodiments of the present invention, the processor 41 is used to execute a weight measurement program stored in the memory 42 and may also implement the following steps: if the matching degree is less than a preset threshold, rotate the second image according to a predetermined direction and a predetermined angle Obtain the third image and calculate the matching degree between the third image and the first image; if the matching degree between the third image and the first image is greater than or equal to the preset threshold, determine the posture of the object to be measured as the matching degree is greater than or equal to the preset The standard posture corresponding to the first image of the threshold; if the matching degree between the third image and the first image is less than the preset threshold, rotate the third image according to the predetermined direction and the predetermined angle; if it rotates according to the predetermined direction and the predetermined angle once, then rotate The matching degree between each subsequent image and the first image is less than a preset threshold, and it is determined that the posture of the object to be measured is an invalid posture.

In other embodiments of the present invention, the processor 41 is used to execute the calculation of the matching degree of the second image and the first image stored in the memory 42 to achieve the following steps: acquiring the corresponding areas of the first image and the second image in the same area The first area; calculate the sum of the area of the first image and the area of the second image to obtain the second area; calculate the ratio of the first area and the second area to obtain the degree of matching between the second image and the first image.

In other embodiments of the present invention, the processor 41 is used to execute the acquisition of image information to be monitored stored in the memory 42 and perform image recognition on the image information to be monitored to obtain the contour of the object to be measured, so as to implement the following steps: obtain a preset A number of objects to be trained that meet predetermined conditions; where the posture of the object to be trained is a standard posture; the number of objects to be trained included on each pixel in the image information to be monitored is obtained and generated based on the number of objects to be trained A first target matrix; generate a second target matrix based on the weight of the object to be trained and the pixel area of the covered pixels; generate a weight mapping matrix based on the second target matrix and the first target matrix.

In other embodiments of the present invention, the processor 41 is configured to execute the acquisition of the number of objects to be trained included in each pixel in the image information to be monitored stored in the memory 42 and generate a first based on the number of objects to be trained The target matrix to achieve the following steps: set the first matrix of M * N; where M * N is the size of the image information to be monitored; sequentially traverse the outline of a preset number of objects to be trained, and The contour is processed to obtain the target part of each object to be trained; the quantity value of the target part of the object to be trained included on each pixel in the image information to be monitored is obtained; according to the correspondence between the pixel and the first matrix, the quantity The value is assigned to the first matrix to obtain a first target matrix; wherein, there is a correspondence between pixels and elements in the first matrix.

In other embodiments of the present invention, the processor 41 is used to execute the pixel area based on the weight of the object to be trained and the covered pixel stored in the memory 42 to generate a second target matrix to achieve the following steps: set M * N The second matrix; measure the weight of each object to be trained, and calculate the pixel area of the pixels covered by the target part of each object to be trained in the image information to be monitored; based on the weight and pixel area of each object to be trained, Calculate the weight weight corresponding to each pixel in the pixels covered by each object to be trained; add the weight weight corresponding to the same pixel in the preset number of pixels covered by the object to be trained to obtain the image information to be monitored The weight value corresponding to each pixel in the pixel; according to the correspondence between the pixel and the second matrix, assign the weight weight to the second matrix to obtain the second target matrix.

In other embodiments of the present invention, the processor 41 is used to execute the calculation of the weight of the object to be measured based on the posture and weight mapping matrix of the object to be measured stored in the memory 42 to achieve the following steps: if the posture of the object to be measured is the standard Posture, obtain the position of the pixel covered by the target part of the object to be measured in the image information to be monitored; generate an M * N mask matrix based on the position of the pixel; calculate the object to be measured based on the mask matrix and the weight mapping matrix Weight.

It should be noted that, for the specific implementation process of the steps performed by the processor in this embodiment, reference may be made to the implementation process in the weight measurement method provided in the embodiments corresponding to FIGS. 1 to 2 and 4, which will not be repeated here.

The weight measuring device provided by the embodiment of the present invention only needs to extract the contour of the object to be measured from the image information of the object to be measured for the object to be measured without human participation, and based on the contour of the object to be measured and pre-training The obtained weight mapping matrix can realize the measurement of the weight of the object to be measured, which solves the problems of large measurement error and high cost when measuring the weight of the pig in the prior art, and realizes the accuracy of the weight of the pig Measurement, and reduces the difficulty of operation and maintenance costs; at the same time, it has universal applicability.

Based on the foregoing embodiments, the embodiments of the present invention provide a computer-readable storage medium that stores one or more programs, and the one or more programs may be executed by one or more processors to implement The following steps: Obtain the image information to be monitored, and perform image recognition on the image information to be monitored to obtain the contour of the object to be measured; determine the posture of the object to be measured based on the contour of the object to be measured; based on the posture of the object to be measured and the weight mapping matrix, calculate the Measure the weight of the subject; among them, the weight mapping matrix is obtained by pre-training.

In other embodiments of the present invention, the one or more programs may be executed by one or more processors to achieve the following steps: acquiring image information to be monitored through an image collector; adopting a specific image segmentation algorithm to perform monitoring on the image information to be monitored Image recognition, determine the contour of the object to be measured from the image information to be monitored.

In other embodiments of the present invention, the one or more programs may be executed by one or more processors to achieve the following steps: acquiring a standard pose for the object to be measured; based on the standard pose and the contour of the object to be measured, determining The posture of the object to be measured.

In other embodiments of the present invention, the one or more programs may be executed by one or more processors to implement the following steps: obtain an image with a standard pose and convert the format of the image with a standard pose to The first image is obtained in a preset format; the image corresponding to the contour of the object to be measured is acquired, and the format of the image corresponding to the contour of the object to be measured is converted into a preset format to obtain the second image; If the degree of matching between the second image and the first image is greater than or equal to a preset threshold, it is determined that the posture of the object to be measured is the standard posture corresponding to the first image whose match degree is greater than or equal to the preset threshold.

In other embodiments of the present invention, the one or more programs may be executed by one or more processors to implement the following steps:

If the matching degree is less than the preset threshold, rotate the second image according to the predetermined direction and the preset angle to obtain the third image, and calculate the matching degree of the third image and the first image;

If the matching degree between the third image and the first image is greater than or equal to a preset threshold, determine that the posture of the object to be measured is a standard posture corresponding to the first image with a matching degree greater than or equal to the preset threshold;

If the matching degree between the third image and the first image is less than a preset threshold, rotate the third image according to a predetermined direction and a predetermined angle;

If the rotation of each image after the rotation in a predetermined direction and a predetermined angle is less than a preset threshold, it is determined that the posture of the object to be measured is an invalid posture.

In other embodiments of the present invention, the one or more programs may be executed by one or more processors to achieve the following steps: acquiring the first area corresponding to the same area in the first image and the second image; calculating the first The sum of the area of the image and the area of the second image yields the second area; the ratio of the first area and the second area is calculated to obtain the degree of matching between the second image and the first image.

In other embodiments of the present invention, the one or more programs may be executed by one or more processors to achieve the following steps: acquiring a preset number of objects to be trained that meet predetermined conditions; acquiring image information to be monitored The number of objects to be trained is included on each pixel of, and a first target matrix is generated based on the number of objects to be trained; a second target matrix is generated based on the weight of the object to be trained and the pixel area of the pixels covered; The second target matrix and the first target matrix generate a weight mapping matrix.

In other embodiments of the present invention, the one or more programs may be executed by one or more processors to achieve the following steps: set a first matrix of M * N; where M * N is the image information to be monitored Size; traverse the outline of a preset number of objects to be trained in sequence, and process the outline of each object to be trained to obtain the target part of each object to be trained; obtain the object to be included on each pixel in the image information to be monitored The quantity value of the target part of the training object; according to the correspondence between the pixels and the first matrix, the quantity value is assigned to the first matrix to obtain the first target matrix; wherein, there is a correspondence between the pixels and the elements in the first matrix relationship.

In other embodiments of the present invention, the one or more programs may be executed by one or more processors to achieve the following steps: setting a second matrix of M * N; measuring the weight of each object to be trained, and calculating The pixel area of the pixels covered by the target part of each object to be trained in the image information to be monitored; based on the weight and pixel area of each object to be trained, the calculation corresponds to each pixel of the pixels covered by each object to be trained Weight weight value; add the weight weight value corresponding to the same pixel among the pixels covered by the preset number of objects to be trained to obtain the weight weight value corresponding to each pixel in the image information to be monitored; Corresponding relationship between the two matrices, weight weights are assigned to the second matrix to obtain the second target matrix.

In other embodiments of the present invention, the one or more programs may be executed by one or more processors to achieve the following steps: acquiring the position of the pixel covered by the object to be measured in the image information to be monitored; based on the pixel The position of the point generates an M * N mask matrix; based on the mask matrix and the weight mapping matrix, the weight of the object to be measured is calculated.

Those skilled in the art should understand that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer usable program code.

The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram and a combination of the flow and / or block in the flowchart and / or block diagram may be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, special-purpose computer, embedded processing machine, or other programmable data processing device to produce a machine that enables the generation of instructions executed by the processor of the computer or other programmable data processing device A device for realizing the functions specified in one block or multiple blocks of one flow or multiple blocks of a flowchart.

These computer program instructions may also be stored in a computer readable memory that can guide a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, the instructions The device implements the functions specified in one block or multiple blocks of the flowchart one flow or multiple flows and / or block diagrams.

These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operating steps are performed on the computer or other programmable device to produce computer-implemented processing, which is executed on the computer or other programmable device The instructions provide steps for implementing the functions specified in one block or multiple blocks of the flowchart one flow or multiple flows and / or block diagrams.

The above are only the preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention.

Industrial applicability

In this embodiment, without manual participation, it is only necessary to extract the contour of the object to be measured from the image information of the object to be monitored, and based on the contour of the object to be measured and the weight mapping matrix obtained by pre-training. The measurement of the weight of the object to be measured is realized, which solves the problems of large measurement error and high cost in measuring the weight of the pig in the prior art, realizes the accurate measurement of the weight of the pig, and reduces the difficulty of operation And maintenance costs; at the same time, it has universal applicability.

Claims

A weight measurement method, characterized in that the method includes:

Acquiring image information to be monitored for the object to be measured, and performing image recognition on the image information to be monitored to obtain the outline of the object to be measured;

Determine the posture of the object to be measured based on the outline of the object to be measured;

Calculate the weight of the object to be measured based on the posture and weight mapping matrix of the object to be measured; wherein the weight mapping matrix is obtained by pre-training.
The method according to claim 1, wherein the acquiring image information to be monitored and performing image recognition on the image information to be monitored to obtain the contour of the object to be measured includes:

Acquiring the image information to be monitored through an image collector;

A specific image segmentation algorithm is used to perform image recognition on the image information to be monitored, and the contour of the object to be measured is determined from the image information to be monitored.
The method according to claim 1, wherein the determining the posture of the object to be measured based on the contour of the object to be measured comprises:

Acquiring a standard pose for the object to be measured;

Based on the standard posture and the contour of the object to be measured, the posture of the object to be measured is determined.
The method according to claim 3, wherein the determining the posture of the object to be measured based on the standard posture and the contour of the object to be measured comprises:

Acquiring an image whose pose is the standard pose, and converting the format of the image whose pose is the standard pose into a preset format to obtain a first image;

Acquiring an image corresponding to the contour of the object to be measured, and converting the format of the image corresponding to the contour of the object to be measured into a preset format to obtain a second image;

Calculating the degree of matching between the second image and the first image;

If the degree of matching between the second image and the first image is greater than or equal to a preset threshold, it is determined that the posture of the object to be measured is a standard posture corresponding to the first image with a degree of matching greater than or equal to the preset threshold.
The method according to claim 4, wherein the method further comprises:

If the matching degree is less than a preset threshold, rotate the second image according to a predetermined direction and a preset angle to obtain a third image, and calculate the matching degree of the third image and the first image;

If the matching degree of the third image and the first image is greater than or equal to a preset threshold, it is determined that the posture of the object to be measured is a standard posture corresponding to the first image with a matching degree greater than or equal to the preset threshold;

If the degree of matching between the third image and the first image is less than a preset threshold, rotate the third image according to the predetermined direction and the predetermined angle;

If after one rotation according to the predetermined direction and predetermined angle, the degree of matching between each rotated image and the first image is less than the preset threshold, it is determined that the posture of the object to be measured is an invalid posture.
The method according to claim 4 or 5, wherein the calculating the degree of matching between the second image and the first image includes:

Acquiring a first area corresponding to the same area in the first image and the second image;

Calculating the sum of the area of the first image and the area of the second image to obtain the second area;

Calculate the ratio of the first area to the second area to obtain the degree of matching between the second image and the first image.
The method according to claim 1, wherein before acquiring the image information to be monitored and performing image recognition on the image information to be monitored to obtain the contour of the object to be measured, the method further includes:

Acquiring a preset number of objects to be trained that meet predetermined conditions; wherein, the posture of the object to be trained is a standard posture;

Acquiring the number of the objects to be trained included on each pixel in the image information to be monitored, and generating a first target matrix based on the number of the objects to be trained;

Generate a second target matrix based on the weight of the object to be trained and the pixel area of the covered pixels;

The weight mapping matrix is generated based on the second target matrix and the first target matrix.
The method according to claim 7, wherein the acquiring the number of the objects to be trained included on each pixel in the image information to be monitored, and generating a number based on the number of the objects to be trained A target matrix, including:

Set a first matrix of M * N; where M * N is the size of the image information to be monitored;

Sequentially traverse the outline of the preset number of objects to be trained, and process the outline of each object to be trained to obtain the target portion of each object to be trained;

Acquiring the quantity value of the target portion of the object to be trained included on each pixel in the image information to be monitored;

According to the correspondence between pixels and the first matrix, assign the quantity value to the first matrix to obtain the first target matrix; wherein, between the pixels and the elements in the first matrix There is a corresponding relationship.
The method according to claim 7, wherein the generating the second target matrix based on the weight of the object to be trained and the pixel area of the covered pixels includes:

Set a second matrix of M * N; where M * N is the size of the image information to be monitored;

Measure the weight of each object to be trained, and calculate the pixel area of pixels covered by the target part of each object to be trained in the image information to be monitored;

Calculate the weight weight corresponding to each pixel in the pixels covered by each object to be trained based on the weight of each object to be trained and the pixel area;

Adding the weight weight corresponding to the same pixel among the pixels covered by the preset number of objects to be trained to obtain the weight weight corresponding to each pixel in the image information to be monitored;

According to the correspondence between pixels and the second matrix, assign the weight weight to the second matrix to obtain a second target matrix.
The method according to any one of claims 7 to 9, wherein the calculating the weight of the object to be measured based on the posture and weight mapping matrix of the object to be measured includes:

If the posture of the object to be measured is a standard posture, obtain the position of the pixel point covered by the target part of the object to be measured in the image information to be monitored;

Generate an M * N mask matrix based on the positions of the pixels;

Based on the mask matrix and the weight mapping matrix, calculate the weight of the object to be measured.
A weight measuring device, characterized in that the device includes: a processor, a memory, and a communication bus;

The communication bus is used to implement a communication connection between the processor and the memory;

The processor is used to execute a weight measurement program stored in the memory to achieve the following steps:

Acquiring image information to be monitored for the object to be measured, and performing image recognition on the image information to be monitored to obtain the outline of the object to be measured;

Determine the posture of the object to be measured based on the outline of the object to be measured;

The weight of the object to be measured is calculated based on the posture and weight mapping matrix of the object to be measured; wherein the weight mapping matrix is obtained by pre-training.
A computer-readable storage medium, characterized in that the computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement claim 1 To any of the steps of the weight measurement method according to any one of 10.