CN110426112A - Live pig weight measuring method and device - Google Patents
Live pig weight measuring method and device Download PDFInfo
- Publication number
- CN110426112A CN110426112A CN201910601442.4A CN201910601442A CN110426112A CN 110426112 A CN110426112 A CN 110426112A CN 201910601442 A CN201910601442 A CN 201910601442A CN 110426112 A CN110426112 A CN 110426112A
- Authority
- CN
- China
- Prior art keywords
- live pig
- measured
- key point
- height
- stern
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01G—WEIGHING
- G01G17/00—Apparatus for or methods of weighing material of special form or property
- G01G17/08—Apparatus for or methods of weighing material of special form or property for weighing livestock
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention provides a live pig weight measuring method and a live pig weight measuring device, which relate to the technical field of artificial intelligence, and the method comprises the following steps: acquiring a video image of a live pig to be measured; inputting the video image into a pre-trained key point detection model to obtain key point thermodynamic diagrams and key point position information of a live pig to be measured, which are output by the model; obtaining a plurality of preset key points according to the matching of the position information of the key points; calculating the hip width, hip height and body length of the live pig to be measured according to preset key points of the live pig to be measured; calculating the volume of the live pig to be measured according to the hip width, the hip height and the body length; and inputting the hip width, hip height, body length and volume into a preset live pig weight regression model to calculate to obtain a weight predicted value of the live pig to be measured. The technical scheme provided by the embodiment of the invention can solve the problem that the accuracy is low when the weight of the live pig is measured by vision in the prior art.
Description
[technical field]
The present invention relates to field of artificial intelligence more particularly to a kind of live pig body weight measurements and device.
[background technique]
Currently, live pig weight be measure live pig whether reach the standard of delivering for sale, and expected revenue is done estimate it is most important
Value carries out live pig weighing using batheroom scale at present, cumbersome, consumes excessive human cost and time cost, especially
It is in large-scale Pig breeding plant, carrying out live pig measured body weight workload one by one will be very huge.And pass through vision measurement live pig
Often accuracy is low for weight.
[summary of the invention]
In view of this, the embodiment of the invention provides a kind of live pig body weight measurement and devices, to solve existing skill
Pass through the weight of the vision measurement live pig often low problem of accuracy in art.
To achieve the goals above, according to an aspect of the invention, there is provided a kind of live pig body weight measurement, described
Method includes: the video image obtained about live pig to be measured；By video image input critical point detection trained in advance
Model obtains the key point thermodynamic chart and key point location information of the live pig to be measured of the model output；According to described
Key point location information matches to obtain multiple default key points；According to the default key point of the live pig to be measured calculate it is described to
Measure hip breadth, the stern height, height of live pig；According to the hip breadth, the body that the stern is high, the height calculates the live pig to be measured
Product；The hip breadth, stern height, the height and the volume are inputted into preset live pig weight regression model, institute is calculated
State the forecast body weight value of live pig to be measured.
Optionally, the meter of the volume that the live pig to be measured is calculated according to the hip breadth, stern height, the height
Calculate formula are as follows: V=W*L*H, wherein W is the hip breadth of the live pig to be measured, and H is that the stern of the live pig to be measured is high, and L is institute
State the height of live pig to be measured.
Optionally, the default key point includes tiptoe, left back elbow, a left side before mouth, head, neck, left front elbow, left front tiptoe, the right side
Afterwards tiptoe, it is right after elbow, it is right after tiptoe, ridge front, chi chung portion, tail, in the middle part of tripe, tripe rear portion；It is described according to the live pig to be measured
Default key point calculate that the hip breadth of the live pig to be measured, stern be high, height, comprising: " head " key point obtained according to matching
The height of live pig to be measured described in the positional information calculation of " tail " key point；According to obtained " chi chung portion " key point of matching and
The stern of live pig to be measured described in the positional information calculation of " tripe in the middle part of " key point is high；" left back elbow " key point obtained according to matching
The hip breadth of live pig to be measured described in the positional information calculation of " right after elbow " key point.
Optionally, in the critical point detection model that video image input is trained in advance, the model is obtained
Before the key point thermodynamic chart and key point location information of the live pig to be measured of output, the method also includes: building institute
State critical point detection model, wherein the hourglass network that the critical point detection model is intensively connected by four is constituted；Using default
Training set the critical point detection model is trained, institute is made using least mean-square error loss function in training process
Hourglass network convergence is stated, the trained critical point detection model is obtained.
Optionally, the training set includes multiple live pig image patterns；The hourglass network includes higher level road and junior road,
The live pig image of the higher level road processing full size, after the junior road is down-sampled to the live pig image progress of the full size again
Carry out a liter sampling processing.
Optionally, described down-sampled to be sampled using maximum pond or average pond, the liter using arest neighbors interpolation method.
Optionally, the hip breadth, stern height, the height and the volume are inputted into preset live pig weight described
Regression model is calculated before the forecast body weight value of the live pig to be measured, the method also includes: acquire several live pigs
The weight and reference data of sample, the reference data include hip breadth, stern height, height and volume；Using the reference data as
Variable, the weight of corresponding live pig sample is as a result, establish live pig weight regression model.
To achieve the goals above, according to an aspect of the invention, there is provided a kind of live pig body weigher, described
Device includes: acquiring unit, for obtaining the video image about live pig to be measured；Input unit is used for the video figure
As input critical point detection model trained in advance, the key point thermodynamic chart of the live pig to be measured of the model output is obtained
With key point location information；Matching unit obtains multiple default key points for matching according to the key point location information；The
One computing unit, for calculating hip breadth, the stern height, body of the live pig to be measured according to the default key point of the live pig to be measured
It is long；Second computing unit, for according to the hip breadth, the volume that the stern is high, the height calculates the live pig to be measured；The
Three computing units return mould for the hip breadth, stern height, the height and the volume to be inputted preset live pig weight
The forecast body weight value of the live pig to be measured is calculated in type.
To achieve the goals above, according to an aspect of the invention, there is provided a kind of computer non-volatile memories are situated between
Matter, the storage medium include the program of storage, control equipment where the storage medium in described program operation and execute
The live pig body weight measurement stated.
In the present solution, utilizing computer vision technique and deep neural network skill as inputting by using video image
The distance between art, the whole body key point of live pig is measured in real time and is tracked, and calculates each key point simultaneously, thus smart
It really calculates about physical parameters such as live pig hip breadth, stern height, height and volumes, utilizes live pig physical parameter known in database
Relationship between weight makes prediction to the exact weight of live pig, improves the accuracy of vision measurement live pig weight.
[Detailed description of the invention]
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this field
For those of ordinary skill, without any creative labor, it can also be obtained according to these attached drawings other attached
Figure.
Fig. 1 is a kind of flow chart of optional live pig body weight measurement provided in an embodiment of the present invention；
Fig. 2 is the schematic diagram of the default key point of live pig provided in an embodiment of the present invention；
Fig. 3 is a kind of schematic diagram of optional live pig body weigher provided in an embodiment of the present invention；
Fig. 4 is a kind of schematic diagram of optional computer equipment provided in an embodiment of the present invention.
[specific embodiment]
For a better understanding of the technical solution of the present invention, being retouched in detail to the embodiment of the present invention with reference to the accompanying drawing
It states.
It will be appreciated that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Base
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts it is all its
Its embodiment, shall fall within the protection scope of the present invention.
The term used in embodiments of the present invention is only to be not intended to be limiting merely for for the purpose of describing particular embodiments
The present invention.In the embodiment of the present invention and the "an" of singular used in the attached claims, " described " and "the"
It is also intended to including most forms, unless the context clearly indicates other meaning.
It should be appreciated that term "and/or" used herein is only a kind of incidence relation for describing affiliated partner, indicate
There may be three kinds of relationships, for example, A and/or B, can indicate: individualism A, exist simultaneously A and B, individualism B these three
Situation.In addition, character "/" herein, typicallys represent the relationship that forward-backward correlation object is a kind of "or".
It will be appreciated that though terminal may be described using term first, second, third, etc. in embodiments of the present invention,
But these terminals should not necessarily be limited by these terms.These terms are only used to for terminal being distinguished from each other out.For example, not departing from the present invention
In the case where scope of embodiments, first terminal can also be referred to as second terminal, and similarly, second terminal can also be referred to as
One terminal.
Depending on context, word as used in this " if " can be construed to " ... when " or " when ...
When " or " in response to determination " or " in response to detection ".Similarly, depend on context, phrase " if it is determined that " or " if detection
(condition or event of statement) " can be construed to " when determining " or " in response to determination " or " when the detection (condition of statement
Or event) when " or " in response to detection (condition or event of statement) ".
Fig. 1 is a kind of flow chart of live pig body weight measurement according to an embodiment of the present invention, as shown in Figure 1, this method
Include:
Step S101 obtains the video image about live pig to be measured.In the present embodiment, video image includes 200
Frame image, in other embodiments, the frame number of video image can carry out adjustment appropriate according to demand.
Video image input critical point detection model trained in advance is obtained the to be measured of model output by step S102
The key point thermodynamic chart and key point location information of live pig.Wherein, the hourglass net that critical point detection model is intensively connected by four
Network is constituted.
Step S103 matches to obtain multiple default key points according to key point location information.
Step S104 calculates the hip breadth of live pig to be measured, stern height, height according to the default key point of live pig to be measured.
Step S105, according to hip breadth, the volume that stern is high, height calculates live pig to be measured.
Step S106 hip breadth, stern height, height and the preset live pig weight regression model of volume input is calculated to be measured
Measure the forecast body weight value of live pig.
In the present solution, utilizing computer vision technique and deep neural network skill as inputting by using video image
The distance between art, the whole body key point of live pig is measured in real time and is tracked, and calculates each key point simultaneously, thus smart
It really calculates about physical parameters such as live pig hip breadth, stern height, height and volumes, utilizes live pig physical parameter known in database
Relationship between weight makes prediction to the exact weight of live pig, improves the accuracy of vision measurement live pig weight.
Optionally, default key point includes tiptoe, left back elbow, left back foot before mouth, head, neck, left front elbow, left front tiptoe, the right side
Point, it is right after elbow, it is right after tiptoe, ridge front, chi chung portion, tail, in the middle part of tripe, tripe rear portion；According to the default key point of live pig to be measured
Calculate hip breadth, the stern height, height of live pig to be measured, comprising:
The height of the positional information calculation live pig to be measured of " head " key point and " tail " key point that are obtained according to matching；Root
The stern of the positional information calculation live pig to be measured of " chi chung portion " key point and " in the middle part of tripe " key point that obtain according to matching is high；According to
Match the hip breadth of the positional information calculation live pig to be measured of obtained " left back elbow " key point and " elbow after right " key point.
With reference to Fig. 2, it is possible to understand that ground, the key point obtained according to matching (head, tail, chi chung portion, in the middle part of tripe, left back elbow and
Elbow behind the right side) location information to calculate, height, stern be high and hip breadth.The weight of live pig and the volume of live pig are directly related, live pig stern
The width in portion and the high sectional area that can be used in indicating live pig.The head of live pig to tail can be used in indicating the height of live pig.
Further, according to hip breadth, the calculation formula for the volume that stern is high, height calculates live pig to be measured are as follows: V=W*L*H,
Wherein, W is the hip breadth of live pig to be measured, and H is that the stern of live pig to be measured is high, and L is the height of live pig to be measured.
It can estimate to obtain the hip breadth, stern height and body of live pig by computer vision technique and deep neural network technology
It is long, to estimate the weight of live pig more accurately.Because average tissue density all day of live pig is roughly equal, live pig body
The volume near-linear dependency of weight and live pig.
Optionally, by hip breadth, stern is high, height and volume input preset live pig weight regression model be calculated it is to be measured
Before the forecast body weight value for measuring live pig, method further include:
The weight and reference data of several live pig samples are acquired, reference data includes hip breadth, stern height, height and volume；
Using reference data as variable, the weight of corresponding live pig sample is as a result, establish live pig weight regression model.
Wherein, the recurrence mode of live pig weight regression model can be linear regression be also possible to lasso trick recurrence, Ke Yili
With the regression coefficient of Least Square Method live pig weight regression model, it is not limited here.Pass through a large amount of live pig sample
The parameters such as height, stern height, hip breadth, volume carry out regression forecasting to live pig weight, and can obtain live pig estimates weight, sentences
Whether medium well pig reaches the standard delivered for sale, simple and fast, it is only necessary to shoot one section of live pig video.In a kind of embodiment
In, camera can be directly installed near the nursings slot on cultivation column to shoot the video of live pig, it can also be by other means
Shooting, it is not limited here.
Optionally, before by video image input critical point detection model trained in advance, method further include: by video
All Image Adjustings in image are presetted pixel；Video image adjusted is input to critical point detection trained in advance
In model.In the present embodiment, video image is first adjusted to 512*256 pixel, to facilitate model to carry out batch processing.
Optionally, in the critical point detection model that video image input is trained in advance, the to be measured of model output is obtained
Before the key point thermodynamic chart and key point location information of live pig, method further include:
Construct critical point detection model, wherein the hourglass network that critical point detection model is intensively connected by four is constituted；
Critical point detection model is trained using preset training set, is damaged in training process using least mean-square error
It loses function and makes hourglass network convergence, obtain trained critical point detection model.
It is to be appreciated that hourglass network can the key point to target object effectively detected, hourglass network includes defeated
Enter layer, convolutional layer, pond layer, up-sampling layer, down-sampling layer etc..When four hourglasses are connected to the network together, previous hourglass
The output of network is the input of an adjacent hourglass network.In order to guarantee the normal update of bottom parameter, each hourglass network is adopted
It is exercised supervision training with relaying supervision strategy to the loss of network.In the present embodiment, it before training starts, needs to initialize
Network parameter, it is 0.00025 that initial learning rate, which is arranged,.
The training set includes multiple live pig image patterns, training before, need to the live pig image pattern in training set into
Row pretreatment, such as live pig image pattern is carried out to be cut to presetted pixel, environmental disturbances region is removed, and to the life after cutting
The each key point of pig image pattern handmarking, wherein the key point n manually marked is expressed as (x, y, z).
Then pretreated training sample is inputted into quadravalence hourglass network, hourglass network includes higher level road and junior road.
Live pig image pattern carry out four times it is down-sampled, every time it is down-sampled before, higher level road handle full size live pig image, junior road pair
A liter sampling processing is carried out again after the live pig image progress of full size is down-sampled.It in the present embodiment, can be from original size
The intermediate characteristic of original size, 1/2,1/4,1/8 is extracted in (512*256), makes figure by rising sampling after extraction feature every time
As restoring to original size, it is added with the data of original size feature, then feature extraction is carried out by a residual error network；
Between down-sampled twice, feature is extracted using three primary modules；Between being added twice, extracted using a primary module special
Sign.
In quadravalence hourglass network, each hourglass network is risen to down-sampled, nearest neighbour interpolation by pond layer
Sampling, thus from top it is lower and the bottom of from and Shangdi can extract crucial point feature in each size.Using jump between hourglass
Connection, so that the key point location information under each resolution ratio preserves.
Above-mentioned liter sampling and down-sampled, in order to help to understand, illustratively, the size of original image is 3*512*256,
Wherein, 512*256 refers to the resolution ratio of RGB image, and 3 indicate feature number of active lanes, the full articulamentum of hourglass network can be with
Feature number of active lanes is directly set, in the present embodiment, feature number of active lanes is 3.
It is down-sampled to refer to the operation of the resolution ratio reduction of image.For example, being carried out to the original image of 3*512*256 primary
Maximum pondization sampling or average pondization sampling, obtain the image of multiple 3*256*108.Liter sampling, which refers to, proposes the resolution ratio of image
The operation risen.For example, obtaining the image of 3*128*128 using arest neighbors interpolation to the image of 3*64*64.
When training, Pn(i, j, k) indicates the prediction possibility of the volume element (i, j, k) of key point n.In order to train mould
The key point n (x, y, z) of type, the mark in training sample is calculated using following formula,
By σ=2, in the training process, use mean square deviation loss as loss function.Specifically, loss function Loss is public
Formula is as follows:
In more figure multi-angles, it will some key point is invisible in figure because of reason is blocked, and exists at this time
When calculating Loss, the corresponding thermodynamic chart of invisible key point will not be counted.
After training, when loss function converges to pre-set interval, critical point detection model, which is meant that, have been trained.
In one embodiment, the key point identified in angles multiple in video image and multiple image is utilized
Location information does three-dimensional reconstruction, so as to obtain absolute coordinate and key point of each key point under the same coordinate system it
Between distance.
Further, the location information of key point is obtained from key point thermodynamic chart, and according to the location information of key point
The distance between two preset key points are calculated, can also use VIO (Vision-Inertial Odometry, vision
Odometer frame) tracking calculates the distance between two key points.
In the present solution, utilizing computer vision technique and deep neural network skill as inputting by using video image
The distance between art, the whole body key point of live pig is measured in real time and is tracked, and calculates each key point simultaneously, thus smart
It really calculates about physical parameters such as live pig hip breadth, stern height, height and volumes, utilizes live pig physical parameter known in database
Relationship between weight makes prediction to the exact weight of live pig, improves the accuracy of vision measurement live pig weight.
The embodiment of the invention provides a kind of live pig body weigher, the device is for executing above-mentioned live pig measured body weight
Method, as shown in figure 3, the device includes: acquiring unit 10, input unit 20, matching unit 30, the first computing unit 40,
Two computing units 50, third computing unit 60.
Acquiring unit 10, for obtaining the video image about live pig to be measured.
Input unit 20 obtains model output for the critical point detection model that video image input is trained in advance
The key point thermodynamic chart and key point location information of live pig to be measured.Wherein, critical point detection model is intensively connected by four
Hourglass network is constituted.
Matching unit 30 obtains multiple default key points for matching according to key point location information.
First computing unit 40, for calculating the hip breadth of live pig to be measured, stern according to the default key point of live pig to be measured
High, height.
Second computing unit 50, for according to hip breadth, the volume that stern is high, height calculates live pig to be measured.
Third computing unit 60, by inputting hip breadth, stern height, height and volume based on preset live pig weight regression model
Calculation obtains the forecast body weight value of live pig to be measured.
In the present solution, utilizing computer vision technique and deep neural network skill as inputting by using video image
The distance between art, the whole body key point of live pig is measured in real time and is tracked, and calculates each key point simultaneously, thus smart
It really calculates about physical parameters such as live pig hip breadth, stern height, height and volumes, utilizes live pig physical parameter known in database
Relationship between weight makes prediction to the exact weight of live pig, improves the accuracy of vision measurement live pig weight.
Optionally, default key point includes tiptoe, left back elbow, left back foot before mouth, head, neck, left front elbow, left front tiptoe, the right side
Point, it is right after elbow, it is right after tiptoe, ridge front, chi chung portion, tail, in the middle part of tripe, tripe rear portion.First computing unit 40 includes the first calculating
Subelement, the second computation subunit, third computation subunit.
First computation subunit, the positional information calculation of " head " key point and " tail " key point for being obtained according to matching
The height of live pig to be measured；Second computation subunit, " chi chung portion " key point and " in the middle part of tripe " pass for being obtained according to matching
The stern of the positional information calculation live pig to be measured of key point is high；Third computation subunit, " left back elbow " for being obtained according to matching
The hip breadth of the positional information calculation live pig to be measured of key point and " elbow after right " key point.
With reference to Fig. 2, it is possible to understand that ground, the key point obtained according to matching (head, tail, chi chung portion, in the middle part of tripe, left back elbow and
Elbow behind the right side) location information to calculate, height, stern be high and hip breadth.The weight of live pig and the volume of live pig are directly related, live pig stern
The width in portion and the high sectional area that can be used in indicating live pig.The head of live pig to tail can be used in indicating the height of live pig.
Further, according to hip breadth, the calculation formula for the volume that stern is high, height calculates live pig to be measured are as follows: V=W*L*H,
Wherein, W is the hip breadth of live pig to be measured, and H is that the stern of live pig to be measured is high, and L is the height of live pig to be measured.
It can estimate to obtain the hip breadth, stern height and body of live pig by computer vision technique and deep neural network technology
It is long, to estimate the weight of live pig more accurately.Because average tissue density all day of live pig is roughly equal, live pig body
The volume near-linear dependency of weight and live pig.
Optionally, device further includes acquisition unit, establishes unit.
Acquisition unit, for acquiring the weight and reference data of several live pig samples, reference data includes hip breadth, stern
High, height and volume；Unit is established, for using reference data as variable, the weight of corresponding live pig sample to be as a result, build
Vertical live pig weight regression model.
Wherein, the recurrence mode of live pig weight regression model can be linear regression be also possible to lasso trick recurrence, Ke Yili
With the regression coefficient of Least Square Method live pig weight regression model, it is not limited here.Pass through a large amount of live pig sample
The parameters such as height, stern height, hip breadth, volume carry out regression forecasting to live pig weight, and can obtain live pig estimates weight, sentences
Whether medium well pig reaches the standard delivered for sale, simple and fast, it is only necessary to shoot one section of live pig video.In a kind of embodiment
In, camera can be directly installed near the nursings slot on cultivation column to shoot the video of live pig, it can also be by other means
Shooting, it is not limited here.
Optionally, device further includes pretreatment unit.
Pretreatment unit, for being presetted pixel by all Image Adjustings in video image；Input unit 20, is also used
In video image adjusted being input in advance trained critical point detection model.In the present embodiment, first by video
Image Adjusting is 512*256 pixel, to facilitate model to carry out batch processing.
Optionally, before by video image input critical point detection model trained in advance, model should first be constructed.Specifically
Ground:
Construct critical point detection model, wherein the hourglass network that critical point detection model is intensively connected by four is constituted；
Critical point detection model is trained using preset training set, is damaged in training process using least mean-square error
It loses function and makes hourglass network convergence, obtain trained critical point detection model.
It is to be appreciated that hourglass network can the key point to target object effectively detected, hourglass network includes defeated
Enter layer, convolutional layer, pond layer, up-sampling layer, down-sampling layer etc..When four hourglasses are connected to the network together, previous hourglass
The output of network is the input of an adjacent hourglass network.In order to guarantee the normal update of bottom parameter, each hourglass network is adopted
It is exercised supervision training with relaying supervision strategy to the loss of network.In the present embodiment, it before training starts, needs to initialize
Network parameter, it is 0.00025 that initial learning rate, which is arranged,.
The training set includes multiple live pig image patterns, training before, need to the live pig image pattern in training set into
Row pretreatment, such as live pig image pattern is carried out to be cut to presetted pixel, environmental disturbances region is removed, and to the life after cutting
The each key point of pig image pattern handmarking, wherein the key point n manually marked is expressed as (x, y, z).
Then pretreated training sample is inputted into quadravalence hourglass network, hourglass network includes higher level road and junior road.
Live pig image pattern carry out four times it is down-sampled, every time it is down-sampled before, higher level road handle full size live pig image, junior road pair
A liter sampling processing is carried out again after the live pig image progress of full size is down-sampled.It in the present embodiment, can be from original size
The intermediate characteristic of original size, 1/2,1/4,1/8 is extracted in (512*256), makes figure by rising sampling after extraction feature every time
As restoring to original size, it is added with the data of original size feature, then feature extraction is carried out by a residual error network；
Between down-sampled twice, feature is extracted using three primary modules；Between being added twice, extracted using a primary module special
Sign.
In quadravalence hourglass network, each hourglass network is risen to down-sampled, nearest neighbour interpolation by pond layer
Sampling, thus from top it is lower and the bottom of from and Shangdi can extract crucial point feature in each size.Using jump between hourglass
Connection, so that the key point location information under each resolution ratio preserves.
Above-mentioned liter sampling and down-sampled, in order to help to understand, illustratively, the size of original image is 3*512*256,
Wherein, 512*256 refers to the resolution ratio of RGB image, and 3 indicate feature number of active lanes, the full articulamentum of hourglass network can be with
Feature number of active lanes is directly set, in the present embodiment, feature number of active lanes is 3.It is down-sampled to refer to the resolution ratio drop of image
Low operation.For example, carrying out primary maximum pondization sampling or average pondization sampling to the original image of 3*512*256, obtain more
The image of a 3*256*108.Sampling is risen to refer to the operation of the increase resolution of image.For example, being adopted to the image of 3*64*64
The image of 3*128*128 is obtained with arest neighbors interpolation.
When training, Pn(i, j, k) indicates the prediction possibility of the volume element (i, j, k) of key point n.In order to train mould
The key point n (x, y, z) of type, the mark in training sample is calculated using following formula,
By σ=2, in the training process, use mean square deviation loss as loss function.Specifically, loss function Loss is public
Formula is as follows:
In more figure multi-angles, it will some key point is invisible in figure because of reason is blocked, and exists at this time
When calculating Loss, the corresponding thermodynamic chart of invisible key point will not be counted.
After training, when loss function converges to pre-set interval, critical point detection model, which is meant that, have been trained.
In one embodiment, the key point identified in angles multiple in video image and multiple image is utilized
Location information does three-dimensional reconstruction, so as to obtain absolute coordinate and key point of each key point under the same coordinate system it
Between distance.
Further, the location information of key point is obtained from key point thermodynamic chart, and according to the location information of key point
The distance between two preset key points are calculated, can also use VIO (Vision-Inertial Odometry, vision
Odometer frame) tracking calculates the distance between two key points.
In the present solution, utilizing computer vision technique and deep neural network skill as inputting by using video image
The distance between art, the whole body key point of live pig is measured in real time and is tracked, and calculates each key point simultaneously, thus smart
It really calculates about physical parameters such as live pig hip breadth, stern height, height and volumes, utilizes live pig physical parameter known in database
Relationship between weight makes prediction to the exact weight of live pig, improves the accuracy of vision measurement live pig weight.
The embodiment of the invention provides a kind of computer non-volatile memory medium, storage medium includes the program of storage,
Wherein, when program is run, equipment where control storage medium executes following steps:
By video image input critical point detection model trained in advance, the described to be measured of the model output is obtained
Measure the key point thermodynamic chart and key point location information of live pig；It is matched to obtain multiple default passes according to the key point location information
Key point；Hip breadth, the stern height, height of the live pig to be measured are calculated according to the default key point of the live pig to be measured；According to institute
State the volume that hip breadth, the stern are high, the height calculates the live pig to be measured；By the hip breadth, stern height, the height
And the volume inputs the forecast body weight value that the live pig to be measured is calculated in preset live pig weight regression model.
Optionally, program run when control storage medium where equipment execute following steps: it is described according to the hip breadth,
The stern is high, the height calculate the live pig to be measured volume calculation formula are as follows: V=W*L*H, wherein W be it is described to
The hip breadth of live pig is measured, H is that the stern of the live pig to be measured is high, and L is the height of the live pig to be measured.
Optionally, when program is run, equipment where control storage medium executes following steps: the default key point packet
Include mouth, head, neck, left front elbow, left front tiptoe, it is right before tiptoe, left back elbow, left back tiptoe, it is right after elbow, it is right after tiptoe, ridge front, ridge
Middle part, tail, tripe middle part, tripe rear portion；The default key point according to the live pig to be measured calculates the live pig to be measured
Hip breadth, stern be high, height, comprising: described in the positional information calculation of " head " key point and " tail " key point that are obtained according to matching to
Measure the height of live pig；According to the positional information calculation of " chi chung portion " key point that matching obtains and " in the middle part of tripe " key point
The stern of live pig to be measured is high；The positional information calculation institute of " left back elbow " key point and " elbow after right " key point that are obtained according to matching
State the hip breadth of live pig to be measured.
Optionally, when program is run, equipment where control storage medium executes following steps: described by the video
Image input critical point detection model trained in advance obtains the key point heating power of the live pig to be measured of the model output
Before figure and key point location information, the method also includes: construct the critical point detection model, wherein the key point
The hourglass network that detection model is intensively connected by four is constituted；The critical point detection model is carried out using preset training set
It trains, the hourglass network convergence is made using least mean-square error loss function in training process, obtain trained described
Critical point detection model.
Optionally, when program is run, equipment where control storage medium executes following steps: the training set includes more
A live pig image pattern；The hourglass network includes higher level road and junior road, and the higher level road handles the live pig image of full size,
A liter sampling processing is carried out again after the junior road is down-sampled to the live pig image progress of the full size.
Optionally, when program is run, equipment where control storage medium executes following steps: described down-sampled using most
Great Chiization or average pond, the liter sampling use arest neighbors interpolation method.
Optionally, when program is run, equipment where control storage medium executes following steps: acquiring several live pig samples
This weight and reference data, the reference data include hip breadth, stern height, height and volume；Using the reference data as change
Amount, the weight of corresponding live pig sample is as a result, establish live pig weight regression model.
Fig. 4 is a kind of schematic diagram of computer equipment provided in an embodiment of the present invention.As shown in figure 4, the meter of the embodiment
Machine equipment 100 is calculated to include: processor 101, memory 102 and storage in the memory 102 and can run on processor 101
Computer program 103, processor 101 execute computer program 103 when realize embodiment in live pig body weight measurement, be
It avoids repeating, not repeat one by one herein.Alternatively, realizing live pig body in embodiment when the computer program is executed by processor 101
The function of each model/unit does not repeat one by one herein in load measurement device to avoid repeating.
Computer equipment 100 can be the calculating such as desktop PC, notebook, palm PC and cloud server and set
It is standby.Computer equipment may include, but be not limited only to, processor 101, memory 102.It will be understood by those skilled in the art that Fig. 3
The only example of computer equipment 100 does not constitute the restriction to computer equipment 100, may include than illustrate it is more or
Less component perhaps combines certain components or different components, such as computer equipment can also be set including input and output
Standby, network access equipment, bus etc..
Alleged processor 101 can be central processing unit (Central Processing Unit, CPU), can also be
Other general processors, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit
(Application Specific Integrated Circuit, ASIC), field programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processor
Deng.
Memory 102 can be the internal storage unit of computer equipment 100, for example, computer equipment 100 hard disk or
Memory.What memory 102 was also possible to be equipped on the External memory equipment of computer equipment 100, such as computer equipment 100 inserts
Connect formula hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash memory
Block (Flash Card) etc..Further, memory 102 can also both including computer equipment 100 internal storage unit or
Including External memory equipment.Memory 102 is for storing other program sum numbers needed for computer program and computer equipment
According to.Memory 102 can be also used for temporarily storing the data that has exported or will export.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In several embodiments provided by the present invention, it should be understood that disclosed system, device and method can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit
It divides, only a kind of logical function partition, there may be another division manner in actual implementation, for example, multiple units or group
Part can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown
Or the mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, device or unit it is indirect
Coupling or communication connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.
The above-mentioned integrated unit being realized in the form of SFU software functional unit can store and computer-readable deposit at one
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
It is each that device (can be personal computer, server or network equipment etc.) or processor (Processor) execute the present invention
The part steps of embodiment the method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (Read-
Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic or disk etc. it is various
It can store the medium of program code.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the present invention.
Claims (10)
1. a kind of live pig body weight measurement, which is characterized in that the described method includes:
Obtain the video image about live pig to be measured；
By video image input critical point detection model trained in advance, the life to be measured of the model output is obtained
The key point thermodynamic chart and key point location information of pig；
It is matched to obtain multiple default key points according to the key point location information；
Hip breadth, the stern height, height of the live pig to be measured are calculated according to the default key point of the live pig to be measured；
According to the hip breadth, the volume that the stern is high, the height calculates the live pig to be measured；
The hip breadth, stern height, the height and the volume are inputted preset live pig weight regression model to be calculated
The forecast body weight value of the live pig to be measured.
2. the method according to claim 1, wherein described according to the hip breadth, stern height, the height meter
Calculate the calculation formula of the volume of the live pig to be measured are as follows:
V=W*L*H, wherein W is the hip breadth of the live pig to be measured, and H is that the stern of the live pig to be measured is high, and L is described to be measured
Measure the height of live pig.
3. according to the method described in claim 2, it is characterized in that, the default key point includes mouth, head, neck, left front elbow, a left side
Preceding tiptoe, it is right before tiptoe, left back elbow, left back tiptoe, it is right after elbow, it is right after tiptoe, ridge front, chi chung portion, tail, in the middle part of tripe, after tripe
Portion；The default key point according to the live pig to be measured calculates the hip breadth of the live pig to be measured, stern height, height, comprising:
The height of the live pig to be measured according to the positional information calculation of obtained " head " key point of matching and " tail " key point；
The live pig to be measured according to the positional information calculation of " chi chung portion " key point that matching obtains and " in the middle part of tripe " key point
Stern it is high；
Live pig to be measured described in the positional information calculation of " left back elbow " key point and " elbow after right " key point that are obtained according to matching
Hip breadth.
4. the method according to claim 1, wherein in the pass that video image input is trained in advance
Key point detection model, obtain the live pig to be measured of model output key point thermodynamic chart and key point location information it
Before, the method also includes:
Construct the critical point detection model, wherein the hourglass network structure that the critical point detection model is intensively connected by four
At；
The critical point detection model is trained using preset training set, is damaged in training process using least mean-square error
It loses function and makes the hourglass network convergence, obtain the trained critical point detection model.
5. according to the method described in claim 4, it is characterized in that, the training set includes multiple live pig image patterns；It is described
Hourglass network includes higher level road and junior road, and the live pig image of the higher level road processing full size, the junior road is to the original
A liter sampling processing is carried out again after the live pig image progress of size is down-sampled.
6. according to the method described in claim 5, it is characterized in that, the maximum pond of the down-sampled use or average pond, institute
Liter sampling is stated using arest neighbors interpolation method.
7. the method according to claim 1, wherein it is described by the hip breadth, the stern is high, the height and
The volume inputs preset live pig weight regression model and is calculated before the forecast body weight value of the live pig to be measured, described
Method further include:
The weight and reference data of several live pig samples are acquired, the reference data includes hip breadth, stern height, height and volume；
Using the reference data as variable, the weight of corresponding live pig sample is as a result, establish live pig weight regression model.
8. a kind of live pig body weigher, which is characterized in that described device includes:
Acquiring unit, for obtaining the video image about live pig to be measured；
Input unit obtains the model output for the critical point detection model that video image input is trained in advance
The live pig to be measured key point thermodynamic chart and key point location information；
Matching unit obtains multiple default key points for matching according to the key point location information；
First computing unit, for calculated according to the default key point of the live pig to be measured the live pig to be measured hip breadth,
Stern height, height；
Second computing unit, for according to the hip breadth, the volume that the stern is high, the height calculates the live pig to be measured；
Third computing unit, for the hip breadth, stern height, the height and the volume to be inputted preset live pig weight
The forecast body weight value of the live pig to be measured is calculated in regression model.
9. device according to claim 8, which is characterized in that described device further include:
Acquisition unit, for acquiring the weight and reference data of several live pig samples, the reference data includes hip breadth, stern
High, height and volume；
Unit is established, for using the reference data as variable, the weight of the corresponding live pig sample to be as a result, establish
Live pig weight regression model.
10. a kind of computer non-volatile memory medium, the storage medium includes the program of storage, which is characterized in that in institute
Equipment perform claim where controlling the storage medium when stating program operation requires live pig weight described in 1 to 7 any one to survey
Amount method.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910601442.4A CN110426112B (en) | 2019-07-04 | 2019-07-04 | Live pig weight measuring method and device |
| PCT/CN2019/104332 WO2021000423A1 (en) | 2019-07-04 | 2019-09-04 | Pig weight measurement method and apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910601442.4A CN110426112B (en) | 2019-07-04 | 2019-07-04 | Live pig weight measuring method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110426112A true CN110426112A (en) | 2019-11-08 |
| CN110426112B CN110426112B (en) | 2022-05-13 |
Family
ID=68408931
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910601442.4A Active CN110426112B (en) | 2019-07-04 | 2019-07-04 | Live pig weight measuring method and device |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110426112B (en) |
| WO (1) | WO2021000423A1 (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111126416A (en) * | 2019-12-12 | 2020-05-08 | 创新奇智(重庆)科技有限公司 | Engine chain wheel identification system and identification method based on key point detection |
| CN111222475A (en) * | 2020-01-09 | 2020-06-02 | 洛阳语音云创新研究院 | Pig tail biting detection method and device and storage medium |
| CN112801118A (en) * | 2021-02-26 | 2021-05-14 | 潘志乐 | Pork pig slaughtering benefit evaluation system and method based on artificial intelligence and big data |
| CN113096178A (en) * | 2021-04-25 | 2021-07-09 | 中国农业大学 | Pig weight estimation method, device, equipment and storage medium |
| CN113344960A (en) * | 2021-05-10 | 2021-09-03 | 河南省农业科学院畜牧兽医研究所 | Live cattle body ruler weight measuring method based on machine vision |
| WO2021196494A1 (en) * | 2020-03-29 | 2021-10-07 | 江苏深农智能科技有限公司 | Livestock and poultry breeding intelligent weighing method and system |
| CN113487107A (en) * | 2021-07-28 | 2021-10-08 | 华南农业大学 | Large animal weight automatic evaluation method, system and medium based on multilayer radial basis network |
| CN113537175A (en) * | 2021-09-16 | 2021-10-22 | 长沙鹏阳信息技术有限公司 | Same-fence swinery average weight estimation method based on computer vision |
| CN113532616A (en) * | 2020-04-15 | 2021-10-22 | 阿里巴巴集团控股有限公司 | Weight estimation method, device and system based on computer vision |
| CN113711992A (en) * | 2021-08-16 | 2021-11-30 | 佛山职业技术学院 | Method and system for acquiring growth state of live pig |
| CN113920453A (en) * | 2021-10-13 | 2022-01-11 | 华南农业大学 | Pig body size weight estimation method based on deep learning |
| CN115331266A (en) * | 2022-10-17 | 2022-11-11 | 天津大学四川创新研究院 | Pig unique identification duplicate removal alarm method |
| CN116206342A (en) * | 2023-04-27 | 2023-06-02 | 广东省农业科学院动物科学研究所 | Pig weight detection method, device, equipment and storage medium |
| CN116519106A (en) * | 2023-06-30 | 2023-08-01 | 中国农业大学 | Method, device, storage medium and equipment for determining weight of live pigs |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113627486B (en) * | 2021-07-12 | 2024-04-16 | 杨龙 | Livestock weight estimation method, device and storage medium |
| CN114926633A (en) * | 2022-03-25 | 2022-08-19 | 成都爱记科技有限公司 | High-precision pig weight estimation method based on deep learning |
| CN116453061B (en) * | 2023-06-08 | 2023-10-03 | 厦门农芯数字科技有限公司 | Remote pig selling supervision method, device and equipment based on image recognition |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103884280A (en) * | 2014-03-14 | 2014-06-25 | 中国农业大学 | Mobile system for monitoring body sizes and weights of pigs in multiple pigsties |
| CN105784083A (en) * | 2016-04-05 | 2016-07-20 | 北京农业信息技术研究中心 | Cow shape measuring method and cow shape measuring system based on stereo vision technology |
| CN206177429U (en) * | 2016-11-04 | 2017-05-17 | 北京农业信息技术研究中心 | Meat chicken weight estimation device based on range image |
| CN108229490A (en) * | 2017-02-23 | 2018-06-29 | 北京市商汤科技开发有限公司 | Critical point detection method, neural network training method, device and electronic equipment |
| CN109141248A (en) * | 2018-07-26 | 2019-01-04 | 深源恒际科技有限公司 | Pig weight measuring method and system based on image |
| CN109579956A (en) * | 2018-12-13 | 2019-04-05 | 北京小龙潜行科技有限公司 | A kind of intelligent limit method for measuring weight, device, electronic equipment and storage medium of raising pigs |
| CN109632058A (en) * | 2018-12-13 | 2019-04-16 | 北京小龙潜行科技有限公司 | A kind of intelligent group rearing method for measuring weight, device, electronic equipment and storage medium of raising pigs |
| CN109632059A (en) * | 2018-12-13 | 2019-04-16 | 北京小龙潜行科技有限公司 | A kind of intelligence method for culturing pigs, system, electronic equipment and storage medium |
| CN109636826A (en) * | 2018-11-13 | 2019-04-16 | 平安科技（深圳）有限公司 | Live pig weight method for measurement, server and computer readable storage medium |
| US20190186981A1 (en) * | 2016-06-03 | 2019-06-20 | Optim Corporation | Animal weight estimation system, animal weight estimation method, and program |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002286421A (en) * | 2001-03-28 | 2002-10-03 | Hideo Minagawa | Apparatus for measuring weight of animal |
| KR20190068266A (en) * | 2017-12-08 | 2019-06-18 | 김정구 | System for measuring weight of livestocks using image analysis and method using the same |
| CN109459119B (en) * | 2018-10-17 | 2020-06-05 | 京东数字科技控股有限公司 | Weight measurement method, device and computer readable storage medium |
| CN109556691A (en) * | 2018-11-05 | 2019-04-02 | 上海睿畜电子科技有限公司 | A kind of method, apparatus and system for estimating field livestock counterpoise |
-
2019
- 2019-07-04 CN CN201910601442.4A patent/CN110426112B/en active Active
- 2019-09-04 WO PCT/CN2019/104332 patent/WO2021000423A1/en active Application Filing
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103884280A (en) * | 2014-03-14 | 2014-06-25 | 中国农业大学 | Mobile system for monitoring body sizes and weights of pigs in multiple pigsties |
| CN105784083A (en) * | 2016-04-05 | 2016-07-20 | 北京农业信息技术研究中心 | Cow shape measuring method and cow shape measuring system based on stereo vision technology |
| US20190186981A1 (en) * | 2016-06-03 | 2019-06-20 | Optim Corporation | Animal weight estimation system, animal weight estimation method, and program |
| CN206177429U (en) * | 2016-11-04 | 2017-05-17 | 北京农业信息技术研究中心 | Meat chicken weight estimation device based on range image |
| CN108229490A (en) * | 2017-02-23 | 2018-06-29 | 北京市商汤科技开发有限公司 | Critical point detection method, neural network training method, device and electronic equipment |
| CN109141248A (en) * | 2018-07-26 | 2019-01-04 | 深源恒际科技有限公司 | Pig weight measuring method and system based on image |
| CN109636826A (en) * | 2018-11-13 | 2019-04-16 | 平安科技（深圳）有限公司 | Live pig weight method for measurement, server and computer readable storage medium |
| CN109579956A (en) * | 2018-12-13 | 2019-04-05 | 北京小龙潜行科技有限公司 | A kind of intelligent limit method for measuring weight, device, electronic equipment and storage medium of raising pigs |
| CN109632058A (en) * | 2018-12-13 | 2019-04-16 | 北京小龙潜行科技有限公司 | A kind of intelligent group rearing method for measuring weight, device, electronic equipment and storage medium of raising pigs |
| CN109632059A (en) * | 2018-12-13 | 2019-04-16 | 北京小龙潜行科技有限公司 | A kind of intelligence method for culturing pigs, system, electronic equipment and storage medium |
Non-Patent Citations (4)
| Title |
|---|
| 付为森等: "种猪体重三维预估模型的研究", 《农业工程学报》 * |
| 李卓 等: "基于机器视觉的猪体质量估测模型比较与优化", 《农业工程学报》 * |
| 李卓 等: "基于深度图像的猪体尺检测系统", 《猪业观察》 * |
| 杨艳等: "种猪体重测量新方法初探", 《畜禽业》 * |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111126416A (en) * | 2019-12-12 | 2020-05-08 | 创新奇智(重庆)科技有限公司 | Engine chain wheel identification system and identification method based on key point detection |
| CN111222475A (en) * | 2020-01-09 | 2020-06-02 | 洛阳语音云创新研究院 | Pig tail biting detection method and device and storage medium |
| CN111222475B (en) * | 2020-01-09 | 2023-05-26 | 洛阳语音云创新研究院 | Pig tail biting detection method, device and storage medium |
| WO2021196494A1 (en) * | 2020-03-29 | 2021-10-07 | 江苏深农智能科技有限公司 | Livestock and poultry breeding intelligent weighing method and system |
| CN113532616A (en) * | 2020-04-15 | 2021-10-22 | 阿里巴巴集团控股有限公司 | Weight estimation method, device and system based on computer vision |
| CN112801118A (en) * | 2021-02-26 | 2021-05-14 | 潘志乐 | Pork pig slaughtering benefit evaluation system and method based on artificial intelligence and big data |
| CN112801118B (en) * | 2021-02-26 | 2024-02-02 | 潘志乐 | Pork pig marketing benefit evaluation system and method based on artificial intelligence and big data |
| CN113096178A (en) * | 2021-04-25 | 2021-07-09 | 中国农业大学 | Pig weight estimation method, device, equipment and storage medium |
| CN113344960A (en) * | 2021-05-10 | 2021-09-03 | 河南省农业科学院畜牧兽医研究所 | Live cattle body ruler weight measuring method based on machine vision |
| CN113344960B (en) * | 2021-05-10 | 2022-11-15 | 河南省农业科学院畜牧兽医研究所 | Live cattle body ruler weight measuring method based on machine vision |
| CN113487107B (en) * | 2021-07-28 | 2024-04-12 | 华南农业大学 | Automatic large animal weight assessment method, system and medium based on multilayer radial basis network |
| CN113487107A (en) * | 2021-07-28 | 2021-10-08 | 华南农业大学 | Large animal weight automatic evaluation method, system and medium based on multilayer radial basis network |
| CN113711992A (en) * | 2021-08-16 | 2021-11-30 | 佛山职业技术学院 | Method and system for acquiring growth state of live pig |
| CN113711992B (en) * | 2021-08-16 | 2022-07-05 | 佛山职业技术学院 | Method and system for acquiring growth state of live pig |
| CN113537175A (en) * | 2021-09-16 | 2021-10-22 | 长沙鹏阳信息技术有限公司 | Same-fence swinery average weight estimation method based on computer vision |
| CN113537175B (en) * | 2021-09-16 | 2022-06-03 | 湖南朗国视觉识别研究院有限公司 | Same-fence swinery average weight estimation method based on computer vision |
| CN113920453A (en) * | 2021-10-13 | 2022-01-11 | 华南农业大学 | Pig body size weight estimation method based on deep learning |
| WO2023060777A1 (en) * | 2021-10-13 | 2023-04-20 | 华南农业大学 | Pig body size and weight estimation method based on deep learning |
| CN115331266A (en) * | 2022-10-17 | 2022-11-11 | 天津大学四川创新研究院 | Pig unique identification duplicate removal alarm method |
| CN116206342A (en) * | 2023-04-27 | 2023-06-02 | 广东省农业科学院动物科学研究所 | Pig weight detection method, device, equipment and storage medium |
| CN116519106A (en) * | 2023-06-30 | 2023-08-01 | 中国农业大学 | Method, device, storage medium and equipment for determining weight of live pigs |
| CN116519106B (en) * | 2023-06-30 | 2023-09-15 | 中国农业大学 | Method, device, storage medium and equipment for determining weight of live pigs |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021000423A1 (en) | 2021-01-07 |
| CN110426112B (en) | 2022-05-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110426112A (en) | Live pig weight measuring method and device | |
| CN109684920B (en) | Object key point positioning method, image processing method, device and storage medium | |
| CN109886388A (en) | A kind of training sample data extending method and device based on variation self-encoding encoder | |
| CN110188720A (en) | A kind of object detection method and system based on convolutional neural networks | |
| CN108875480A (en) | A kind of method for tracing of face characteristic information, apparatus and system | |
| Rekaya et al. | Bayesian analysis of lactation curves of Holstein-Friesian cattle using a nonlinear model | |
| CN108229555A (en) | Sample weights distribution method, model training method, electronic equipment and storage medium | |
| CN109902546A (en) | Face identification method, device and computer-readable medium | |
| CN109446889A (en) | Object tracking method and device based on twin matching network | |
| CN109141248A (en) | Pig weight measuring method and system based on image | |
| CN110363086A (en) | Diagram data recognition methods, device, computer equipment and storage medium | |
| CN112735094A (en) | Geological disaster prediction method and device based on machine learning and electronic equipment | |
| CN105912991A (en) | Behavior identification method based on 3D point cloud and key bone nodes | |
| CN112926541A (en) | Sleeping post detection method and device and related equipment | |
| Junttila et al. | Sparse Bayesian estimation of forest stand characteristics from airborne laser scanning | |
| CN110263731A (en) | A kind of single step face detection system | |
| CN112348909A (en) | Target positioning method, device, equipment and storage medium | |
| CN107918688A (en) | Model of place method for dynamic estimation, data analysing method and device, electronic equipment | |
| Ormerod | Skew-normal variational approximations for Bayesian inference | |
| CN110163103A (en) | A kind of live pig Activity recognition method and apparatus based on video image | |
| CN116519106B (en) | Method, device, storage medium and equipment for determining weight of live pigs | |
| CN112966547A (en) | Neural network-based gas field abnormal behavior recognition early warning method, system, terminal and storage medium | |
| CN109871814A (en) | Estimation method, device, electronic equipment and the computer storage medium at age | |
| CN109410046A (en) | Target stock selection method, device and storage medium | |
| CN108846341A (en) | A kind of remote sensing images lake ice classifying identification method neural network based |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |