CN113487107B - Automatic large animal weight assessment method, system and medium based on multilayer radial basis network - Google Patents
Automatic large animal weight assessment method, system and medium based on multilayer radial basis network Download PDFInfo
- Publication number
- CN113487107B CN113487107B CN202110854716.8A CN202110854716A CN113487107B CN 113487107 B CN113487107 B CN 113487107B CN 202110854716 A CN202110854716 A CN 202110854716A CN 113487107 B CN113487107 B CN 113487107B
- Authority
- CN
- China
- Prior art keywords
- layer
- radial basis
- weight
- radial
- network model
- 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.)
- Active
Links
- 241001465754 Metazoa Species 0.000 title claims abstract description 85
- 238000000034 method Methods 0.000 title claims abstract description 34
- 210000002569 neuron Anatomy 0.000 claims abstract description 48
- 230000006870 function Effects 0.000 claims description 72
- 238000012549 training Methods 0.000 claims description 25
- 230000037396 body weight Effects 0.000 claims description 24
- 238000009795 derivation Methods 0.000 claims description 21
- 238000010606 normalization Methods 0.000 claims description 17
- 239000011159 matrix material Substances 0.000 claims description 15
- 238000013507 mapping Methods 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 6
- 230000009471 action Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 230000010473 stable expression Effects 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 2
- 230000003187 abdominal effect Effects 0.000 claims 1
- 210000005036 nerve Anatomy 0.000 claims 1
- 230000036544 posture Effects 0.000 abstract 1
- 238000013528 artificial neural network Methods 0.000 description 20
- 241000282887 Suidae Species 0.000 description 8
- 230000000694 effects Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 4
- 238000005259 measurement Methods 0.000 description 4
- 238000000691 measurement method Methods 0.000 description 3
- 210000001015 abdomen Anatomy 0.000 description 2
- 238000003491 array Methods 0.000 description 2
- 241000283690 Bos taurus Species 0.000 description 1
- 241001494479 Pecora Species 0.000 description 1
- 206010042434 Sudden death Diseases 0.000 description 1
- 241000700605 Viruses Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000013011 mating Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035764 nutrition Effects 0.000 description 1
- 235000016709 nutrition Nutrition 0.000 description 1
- 230000008929 regeneration Effects 0.000 description 1
- 238000011069 regeneration method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000003938 response to stress Effects 0.000 description 1
- 229940125723 sedative agent Drugs 0.000 description 1
- 239000000932 sedative agent Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
Abstract
The invention discloses a method, a system and a medium for automatically evaluating the weight of a large animal based on a multilayer radial basis network, wherein the method comprises the following steps: each layer randomly generates radial basis function center points, regenerates Cheng Jingxiang base neurons to form radial base layers, and finally connects the radial base layers and the full-connection layers to construct a multi-layer radial base network model; inputting normalized large animal body condition parameters to train a multi-layer radial basis network model, and optimizing parameters of the model by adopting a loss function; and (5) carrying out automatic weight estimation by using the trained multilayer radial basis network model. The method adopts one-dimensional large animal body condition parameters to predict the weight of the animal, is basically not influenced by animal postures and movements, and has high robustness for estimating the animal weight; the invention connects the radial base network layers through the full connection layer, has higher fitting capacity and higher estimation precision only under the condition of inputting one-dimensional body condition.
Description
Technical Field
The invention belongs to the technical field of deep learning, and particularly relates to a method, a system and a medium for automatically evaluating the weight of a large animal based on a multi-layer radial basis network.
Background
The living weight is an important index of the large animals, the weight of the large animals can be timely and accurately obtained, the growth condition and the health condition of the large animals can be immediately evaluated, the feed absorptivity of the large animals can be detected according to the weight, or the large animals with different nutrition conditions can be respectively fed, so that the maximum utilization rate and the optimal growth control of the feed can be achieved. How to automatically and accurately measure the weight of a large animal is one of the important research points of the current intelligent agriculture. The traditional large animal weight detection usually needs to be in direct contact with the large animal, the large animal needs to be moved to weighing equipment such as an electronic scale, the whole process is time-consuming and labor-consuming, the error of manual measurement is large, and epidemic disease propagation is easy to cause due to human and animal contact; sometimes, the use of sedatives and other medicaments is also needed to assist, so that great pressure is brought to large animals, daily activities such as feeding, mating and the like are affected, and sudden death of the large animals is even caused, so that great economic loss is caused.
Recently, various methods for evaluating body weight of large animals have been proposed, which are mainly classified into 2 major categories, i.e., direct measurement and indirect measurement. In the direct measurement method, downward force other than the body weight caused by movement of a large animal or the like is hardly solved. In the indirect measurement method, the two-dimensional area of the large animal is required to be obtained in the current method, the acquisition of the parameter has high precision requirement on the captured image, the large animal is required to be in a relatively fixed position, and the large animal is extremely easy to influence by the posture of the large animal. According to the principle of animal feeding, potential relationship exists between the body condition and the weight of the large animal, and when the variety of the large animal is determined, the automatic estimation of the weight of the large animal can be realized by constructing a deduction model of the body condition and the weight. Meanwhile, if only a plurality of one-dimensional body condition characteristics, such as height, girth and the like, of the large animals can be acquired, the requirements for capturing images of the large animals are greatly reduced, and the captured images have high robustness.
Disclosure of Invention
The invention aims to overcome the defects and shortcomings of the prior art and provide a method, a system and a medium for automatically evaluating the weight of a large animal based on a multi-layer radial base network.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
the invention provides a large animal weight automatic assessment method based on a multilayer radial basis network, which comprises the following steps:
generating center points in a random position mode according to the number of the set center points of each radial base layer, generating corresponding radial basis functions and radial basis neurons based on the center points, adjusting dimensions by connecting all the connecting layers to connect the next layer, and repeating the steps to construct a multi-layer radial basis network model;
training a multi-layer radial basis network model, namely inputting body condition parameters subjected to normalization processing into the multi-layer radial basis network model to obtain output data, inputting the output data and standard data as parameters of a loss function to calculate a loss value, optimizing the parameters of the multi-layer radial basis network model by using the loss function, and obtaining the trained multi-layer radial basis network model when the loss value stably reaches a certain expected minimum value;
and taking the normalized body condition parameter result of the large animal to be predicted as input data, and inputting the input data into a trained multilayer radial base network model, wherein the output of the multilayer radial base network model is the estimated weight value of the large animal to be predicted.
As a preferable technical scheme, the central point is a vector with the same dimension as the input data, represents a point in the dimension space, is random in initial position, and is updated and adjusted through loss function derivation subsequently;
the process of generating the radial basis function corresponding to the center point and the radial basis neuron specifically comprises the following steps: the radial basis neuron control number sigma is randomly generated by each central point, and the radial basis neurons are regenerated by each central pointWherein (1)>Is the ith center point of the kth layer, x k For the k-th layer input data, sigma controls radial basis functionThe initial value of the warp element is random, and then updating and adjusting are carried out through loss function derivation; the output data of the layer is a matrix with the number of lines being the number of neurons of the radial base network of the layer and the number of columns being 1;
the full connection layer is a matrix with the row number equal to the number of output matrix columns of the radial base layer, and the dimension of output data is adjusted by matrix multiplication.
As a preferable technical scheme, the working process of the radial basis function is as follows:
mapping the input data from low latitude to high latitude, and sequentially calculating Euclidean distance between the data mapped to high dimension and each center point of the kth layerNamely +.>Wherein x is k C for mapping the input data on all radial basis cores of the kth layer i C is the ith center point i And x i Is a one-dimensional vector with the same dimension;
shape of the kth layer ith radial basis function controlled according to sigmaAnd by means of a radial basis function, i.e.>Wherein->Sequentially outputting the reaction values of all the center points under the action of the radial basis function for the output value of the ith radial basis function of the kth layer;
connecting the full connection layer and obtaining its outputWherein->B is the output value of the jth neuron in the kth layer in the fully connected layer k Represents the bias number of the k layer of the fully connected layer, ">The weight of the j-th fully connected neuron in the k-th layer is represented, and then the radial basis network of the next layer is connected.
As a preferable technical scheme, the body condition parameters comprise body length, body height, body width, abdomen circumference and waistline.
As an preferable technical scheme, the normalization processing specifically includes respectively normalizing the input body condition parameters and the corresponding body weight, where the normalization methods are the same, and specifically has the following formula:
wherein x and x' respectively represent a one-dimensional vector formed by body condition parameters of a certain head large animal before normalization and a one-dimensional vector formed by body condition parameters of the head large animal after normalization, min is a one-dimensional vector formed by a theoretical minimum value of each body condition of the large animal, and max is a one-dimensional vector formed by a theoretical maximum value of each body condition of the large animal.
As an preferable technical scheme, the step of inputting the normalized body condition parameters into the multi-layer radial basis network model to obtain output data specifically includes the following steps:
calculating the ith input data of the kth layer and the center point of each neuron of the kth layerEuropean distance->
The data is expanded by power series:mapping from low dimension to high dimension in an expanded form of polynomial kernels to extract more feature information, where n k Representing the number of neurons in the k-th layer, x k Input data representing the kth layer, +.>Representing the kth layer ith neuron center point;
the dimension of the output data matrix is adjusted by the last full-connection layer, the estimated weight y' of the current multi-layer radial basis network model is output, and inverse normalization is carried out: y=y' (Weight) max -Weight min )+Weight min Obtaining an estimated body Weight of the large animal, wherein Weight is max And Weight min Theoretical body weight maximum and minimum for this large animal.
As a preferable technical solution, the loss function is specifically:
the function is expressed by expansion, and the general formula of the obtained network is as follows:
wherein L is a loss value obtained by a loss function, Y i For pre-measured standard weight data of the ith large animal, y i Weight data of the ith large animal estimated for the multi-layer radial basis network, N is the data amount input into the multi-layer radial basis network model, N k Representing the number of neurons of the k-th layer network, w k Full connection layer weight, x, representing a k-layer network k Representing the input of a layer k network,representing the center point of the jth neuron of the kth layer.
As an optimal technical scheme, the parameters for optimizing the multi-layer radial basis network model by using the loss function are specifically as follows:
firstly, aiming at a loss function, taking a central point c of a radial base network as a variable to conduct gradient derivation, and subtracting a value obtained by derivation from the central point before optimization to realize updating of central point parameters;
then, aiming at the loss function, taking the radial basis neuron control number sigma as a variable to conduct gradient derivation, and subtracting the value obtained by the derivation from the sigma to conduct parameter updating;
finally, aiming at the loss function, carrying out gradient derivation by taking the full-connection layer neuron weight w as a variable, and subtracting a value obtained by derivation from the full-connection layer neuron weight to carry out parameter updating;
and continuously inputting a plurality of groups of large animal normalized body condition parameters and body weight for iteration for multiple times, and updating the central point, the neuron control number and the weight of the full-connection layer until a nonlinear model of stable expression body condition and body weight is obtained.
The invention further provides a large animal weight automatic assessment system based on the multilayer radial basis network, which is applied to the large animal weight automatic assessment method based on the multilayer radial basis network, and comprises a construction module, a training module and a prediction module;
the construction module is used for generating center points in a random position mode according to the number of the set center points of each radial base layer, generating corresponding radial basis functions based on the center points, further obtaining radial basis neurons to form the radial base layers, connecting the radial base layers with the full-connection layers to generate a radial basis network model, and repeating the steps to construct the multi-layer radial basis network model;
the training module is used for training the multi-layer radial base network model, firstly inputting the normalized body condition parameters into the multi-layer radial base network model to obtain output data, inputting the output data and the standard data as parameters of a loss function to calculate a loss value, then optimizing the parameters of the multi-layer radial base network model by using the loss function, and obtaining the trained multi-layer radial base network model when the loss value is stable and reaches a certain expected minimum value;
the prediction module is used for taking the normalized body condition parameters of the large animal to be predicted as input, inputting the normalized body condition parameters into the trained multilayer radial basis network model, calculating a normalized body weight prediction result, and then carrying out inverse normalization on the result, wherein the output of the multilayer radial basis network model is used for obtaining the estimated body weight value of the large animal to be predicted.
In a further aspect, the present invention provides a computer readable storage medium storing a program which, when executed by a processor, implements the method for automatically assessing body weight of a large animal based on a multi-layer radial basis network.
Compared with the prior art, the invention has the following advantages and beneficial effects:
(1) Compared with the two-dimensional area data and the three-dimensional volume data of the large animal, the method has the advantages that the influence of the gesture and the movement of the large animal is basically avoided, and the influence of the gesture and the movement on the two-dimensional and the three-dimensional data is larger, so that the method has larger robustness for estimating the weight of the large animal.
(2) The network model constructed by the invention adopts a Gaussian radial basis function, wherein the local fitting property enables the Gaussian kernel to influence a certain local area of an input space only by a few weights, so that the data input each time only needs to be adjusted for the corresponding parameters, and the learning speed is higher. If the radial basis network layer constructed by the method has enough Gaussian kernel functions which are uniformly distributed, any function of interest can be approximated theoretically.
(3) According to the method, the multi-layer radial base network layers are connected through the full connection layer, the residual function of the front-layer radial base network is fitted by using the radial base network of the rear layer in sequence, and the multi-layer radial base network with higher fitting capacity is obtained in sequence, so that the estimation accuracy is higher only under the condition of inputting one-dimensional body condition.
(4) The invention solves the problems of the prior weight measurement method mainly based on manual measurement in large animal cultivation, reduces human and animal contact, avoids animal stress response caused by human and animal contact, improves animal welfare, reduces the risk of virus transmission between human and animal, and provides a non-contact automatic weight assessment method.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for automatically assessing weight of a live pig based on a multi-layer radial basis network in accordance with an embodiment of the present invention;
FIG. 2 is a framework diagram of a multi-layer radial basis network model in accordance with an embodiment of the present invention;
FIG. 3 (a) is a graph showing the effect of sow samples on weight estimation after multi-layer radial basis network training;
FIG. 3 (b) is a graph showing the effect of boar samples on weight estimation after multi-layer radial basis network training;
FIG. 3 (c) is a graph showing the effect of a mixed boar and sow sample on weight estimation after multi-layer radial basis network training;
FIG. 4 (a) is a graph showing the effect of sow samples on weight estimation after training on a fully-connected neural network;
FIG. 4 (b) is a graph showing the effect of a boar sample on weight estimation after training on a fully connected neural network;
FIG. 4 (c) is a graph showing the effect of a mixed boar and sow sample on weight estimation after training on a fully connected neural network;
FIG. 5 (a) is a graph of the number of iterations versus loss value for a fully connected neural network;
FIG. 5 (b) is a graph of multi-layer radial basis network iteration number versus loss value;
FIG. 6 is a schematic structural diagram of an automatic live pig weight assessment system based on a multilayer radial basis network according to an embodiment of the present invention;
fig. 7 is a schematic structural diagram of a storage medium according to an embodiment of the present invention.
Detailed Description
In order to enable those skilled in the art to better understand the present application, the following description will make clear and complete descriptions of the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. It will be apparent that the described embodiments are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.
Reference in the specification to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly understand that the embodiments described herein may be combined with other embodiments.
As shown in fig. 1, this embodiment provides a method for automatically evaluating the weight of a large animal based on a multi-layer radial basis network, and this embodiment uses live pigs as an example, and other large animals such as cattle, sheep, ma Dengtong are suitable for the technical scheme of this embodiment, and the method for automatically evaluating the weight of a live pig according to this embodiment includes the following steps:
(1) A stage of constructing a multi-layer radial basis network model, comprising:
(1-1) setting the number of center points of each layer, and randomly generating a corresponding number of center points c; the center point c is a vector with the same dimension as the input data, represents a point in the dimension space, is random in initial position, and is updated and adjusted through loss function derivation.
(1-2) random generation of radial basis neuron control number σ per center point, regeneration of radial basis neurons per center pointWherein (1)>Is the ith center point of the kth layer, x k Inputting data for a kth layer, sigma controlling the shape and the action range of radial basis neurons, and carrying out updating adjustment by leading a loss function; the output data of the layer is a matrix with the number of lines being the number of neurons of the radial base network of the layer and the number of columns being 1;
(1-3) connecting the full connection layer for dimension adjustment to connect the next radial base network. The full connection layer is a matrix with the row number equal to the number of output matrix columns of the radial base layer, and the dimension of output data is adjusted by matrix multiplication.
(1-4) repeating the steps (1-1) to (1-3) to construct a multilayer radial basis network model.
(2) A training multi-layer radial basis network model stage comprising:
(2-1) normalizing the body condition parameters of the live pigs, wherein the body condition parameters are represented by the following formula:
wherein x and x' represent body condition parameters before and after normalization, respectively, and the body condition parameters comprise body length, body height, body width, abdomen circumference and waistline; min is the minimum value of the weight of the theoretical live pigs, and max is the maximum value of the weight of the theoretical live pigs. The body condition parameters of the live pigs are normalized to be in the range of 0-1, so that the study of the multilayer radial basis network is facilitated.
(2-2) inputting the normalized body condition parameters into a multi-layer radial basis network model to obtain output data of the model, wherein the output data are specifically as follows:
calculating the ith input data of the kth layer and the center point of each neuron of the kth layerEuropean distance->
The data is expanded by power series:mapping from low dimension to high dimension in an expanded form of polynomial kernels to extract more feature information, where n k Representing the number of neurons in the k-th layer, x k Input data representing the kth layer, +.>Representing the kth layer ith neuron center point;
the dimension of the output data matrix is adjusted by the last full-connection layer, the estimated weight y' of the current multi-layer radial basis network model is output, and inverse normalization is carried out: y=y' (Weight) max -Weight min )+Weight min Obtaining an estimated Weight of a live pig, wherein Weight is Weight max And Weight min Maximum and minimum theoretical body weight for the pig.
(2-3) inputting the output data and the standard data into a loss function, performing gradient derivation based on the loss function, and updating parameters of the multilayer radial basis network model;
the loss function is specifically:
the function is expressed by expansion, and the general formula of the obtained network is as follows:
wherein L is the loss obtained by the loss functionValue of Y i For the pre-measured standard weight data of the ith live pig, y i Weight data of the ith live pig estimated for the multilayer radial basis network, N is the data amount input into the multilayer radial basis network model, N k Representing the number of neurons of the k-th layer network, w k Full connection layer weight, x, representing a k-layer network k Representing the input of a layer k network,representing the center point of the jth neuron of the kth layer.
The parameters for updating the multilayer radial base network model are specifically as follows:
gradient derivation is carried out by taking the central point c as a variable through a loss function, and the value obtained by subtracting the derivation from the central point c is used for parameter updating;
and performing gradient derivation by using the radial basis neuron control number sigma as a variable through a loss function, and subtracting the derived value from the sigma to perform parameter updating.
And carrying out gradient derivation by taking the full-connection layer neuron weight as a variable through a loss function, and subtracting the derived value from the full-connection layer neuron weight to carry out parameter updating.
(2-4) repeating the steps (2-1) to (2-3), and obtaining the nonlinear model of stable expression body condition and body weight after iterative updating.
(3) Predicting by using the trained multilayer radial base network model;
(3-1) normalizing the body condition parameters of the live pigs to be predicted;
(3-2) inputting the normalized body condition parameters of the live pigs into a multi-layer radial basis network model;
and (3-3) automatically obtaining the estimated weight value of the live pig to be predicted by the multi-layer radial basis network model.
FIG. 2 is a framework diagram of a multi-layer radial basis network model of the present invention depicting Radial Basis Function (RBF) kernel functions of a first layer. Firstly, mapping input data from low latitude to high latitude, and sequentially calculating Euclidean distance between the data mapped to the high latitude and each center point of a kth layerNamely +.>Wherein x is k C for mapping the input data on all radial basis cores of the kth layer i C is the ith center point i And x i Is a one-dimensional vector with the same dimension;
shape of the kth layer ith radial basis function controlled according to sigmaAnd by means of a radial basis function, i.e.>Wherein->Sequentially outputting the reaction values of all the center points under the action of the radial basis function for the output value of the ith radial basis function of the kth layer;
connecting the full connection layer and obtaining its outputWherein->B is the output value of the jth neuron in the kth layer in the fully connected layer k Represents the bias number of the k layer of the fully connected layer, ">The weight of the j-th fully-connected neuron in the k-th layer is represented, then the next radial base network is connected until the weight estimated value is finally output, and each RBF layer is connected together through the fully-connected layer, so that the multi-layer radial base network used by the invention is formed.
Fig. 3 (a), fig. 3 (b), and fig. 3 (c) respectively describe the predicted results of the multi-layer radial basis network (multi-layer RBF network) on the body weight of the boar, the sow, and the boar after the network training, and fig. 4 (a), fig. 4 (b), and fig. 4 (c) respectively describe the predicted results of the fully-connected neural network (BP network) on the body weight of the boar, the sow, and the boar after the mixed network training. In the graph, the abscissa is the actual weight of a live pig, the ordinate is the predicted weight of a neural network on the live pig, each point represents each pig, the oblique dotted line represents perfect weight prediction, so the closer to the point of the oblique dotted line, the more perfect the prediction result is shown, the more the weight of a sow (fig. 3 (a)) is predicted to be better than that of a boar (fig. 3 (b)) in the multilayer radial base network, the weight prediction of the boar is better than that of a boar and sow (fig. 3 (c)), the weight prediction effect of the multilayer radial base network is better than that of a fully connected neural network as a whole, and the better performance of the local fitting of the multilayer radial base network on the body condition-weight prediction is proved.
Fig. 5 (a) and fig. 5 (b) describe a loss curve of a training set and a loss curve of a test set of the multi-layer radial basis function neural network and the fully connected neural network in network training, and it can be seen that the training iteration number of the multi-layer radial basis function neural network is almost completed in about one hundred times, the training iteration number of the fully connected neural network needs more than three hundred times, and the learning speed of the multi-layer radial basis function neural network is higher than that of the fully connected neural network.
As can be seen from table 1, the prediction result of the multi-layer radial base network of the present embodiment is superior to that of the fully connected neural network; wherein, the absolute error is defined as the average value of the difference between the predicted weight and the actual weight, which can directly reflect the error of the weight prediction; the relative error is defined as the average error value compared with the average weight of the live pigs, and can scale all the errors to the same scale and reflect the accuracy of the errors.
| Sex (sex) | Network model | Absolute error of | Relative error |
| Duroc boar | Radial basis neural network | 1.71 | 1.10％ |
| Duroc sow | Radial basis neural network | 1.42 | 1.02％ |
| Duroc male and female mixing | Radial basis neural network | 1.95 | 1.05％ |
| Duroc boar | Fully connected neural network | 1.96 | 1.70％ |
| Duroc sow | Fully connected neural network | 1.85 | 1.35％ |
| Duroc male and female mixing | Fully connected neural network | 2.2 | 1.95％ |
Table 1 error statistics of multi-layer radial basis network and BP network predictions
In another embodiment of the present application, as shown in fig. 6, there is provided an automatic pig weight assessment system based on a multi-layer radial basis network, the system comprising a construction module, a training module, and a prediction module;
the construction module is used for generating center points in a random position mode according to the number of the set center points of each radial base layer, generating corresponding radial basis functions and radial basis neurons based on the center points, adjusting dimensions by connecting all the connecting layers to connect the next layer, and repeating the steps to construct a multi-layer radial basis network model;
the training module is used for training the multi-layer radial base network model, firstly inputting the normalized body condition parameters into the multi-layer radial base network model to obtain output data, constructing a loss function according to the output data and standard data, and then optimizing the parameters of the multi-layer radial base network model by using the loss function to obtain a trained multi-layer radial base network model;
the prediction module is used for taking the normalized body condition parameter result of the live pig to be predicted as input, inputting the normalized body condition parameter result into a trained multilayer radial basis network model, and outputting the multilayer radial basis network model to obtain a weight estimated value of the large animal to be predicted.
It should be noted that, the system provided in the above embodiment is only exemplified by the division of the above functional modules, and in practical application, the above functional allocation may be performed by different functional modules according to needs, that is, the internal structure is divided into different functional modules to perform all or part of the functions described above, and the system is the automatic assessment method for the body weight of the large animal based on the multilayer radial base network applied in the above embodiment.
As shown in fig. 7, in another embodiment of the present application, there is further provided a computer readable storage medium storing a program, where the program when executed by a processor implements a method for automatically evaluating the weight of a large animal based on a multi-layer radial basis network, specifically:
generating center points in a random position mode according to the number of the set center points of each radial base layer, generating corresponding radial basis functions and radial basis neurons based on the center points, adjusting dimensions by connecting all the connecting layers to connect the next layer, and repeating the steps to construct a multi-layer radial basis network model;
training a multi-layer radial basis network model, firstly inputting the normalized body condition parameters into the multi-layer radial basis network model to obtain output data, constructing a loss function according to the output data and standard data, and then optimizing the parameters of the multi-layer radial basis network model by using the loss function to obtain a trained multi-layer radial basis network model;
and taking the normalized body condition parameter result of the live pig to be predicted as input, and inputting the normalized body condition parameter result into a trained multilayer radial basis network model, wherein the output of the multilayer radial basis network model is the estimated weight value of the large animal to be predicted.
It is to be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above examples, and any other changes, modifications, substitutions, combinations, and simplifications that do not depart from the spirit and principle of the present invention should be made in the equivalent manner, and the embodiments are included in the protection scope of the present invention.
Claims (5)
1. The automatic large animal weight assessment method based on the multilayer radial basis network is characterized by comprising the following steps of:
generating center points in a random position mode according to the number of the set center points of each radial base layer, generating corresponding radial basis functions and radial basis neurons based on the center points, adjusting dimensions by connecting all the connecting layers to connect the next layer, and repeating the steps to construct a multi-layer radial basis network model; the center point is a vector with the same dimension as the input data, represents a point in the dimension space, is random in initial position, and is updated and adjusted through loss function derivation;
the process of generating the radial basis function corresponding to the center point and the radial basis neuron specifically comprises the following steps: the radial basis neuron control number sigma is randomly generated by each central point, and the radial basis neurons are regenerated by each central pointWherein (1)>Is the ith center point of the kth layer, x k Inputting data for a kth layer, sigma controlling the shape and the action range of radial basis neurons, and carrying out updating adjustment by leading a loss function; the output data of the layer is a matrix with the number of lines being the number of neurons of the radial base network of the layer and the number of columns being 1;
the full connection layer is a matrix with the row number equal to the number of columns of the radial base layer output matrix, and the dimension of output data is adjusted by matrix multiplication;
the working process of the radial basis function is as follows:
mapping the input data from low latitude to high latitude, and sequentially calculating Euclidean distance between the data mapped to high dimension and each center point of the kth layerNamely +.>Wherein x is k C for mapping the input data on all radial basis cores of the kth layer i C is the ith center point i And x i Is a one-dimensional vector with the same dimension;
shape of the kth layer ith radial basis function controlled according to sigmaAnd by means of a radial basis function, i.e.>Wherein->Sequentially outputting the reaction values of all the center points under the action of the radial basis function for the output value of the ith radial basis function of the kth layer;
connecting the full connection layer and obtaining its outputWherein->B is the output value of the jth neuron in the kth layer in the fully connected layer k Represents the bias number of the k layer of the fully connected layer, ">Representing the weight of the j-th fully-connected neuron in the k-th layer, and then connecting the next radial basis network;
training a multi-layer radial basis network model, namely inputting body condition parameters subjected to normalization processing into the multi-layer radial basis network model to obtain output data, inputting the output data and standard data as parameters of a loss function to calculate a loss value, optimizing the parameters of the multi-layer radial basis network model by using the loss function, and obtaining the trained multi-layer radial basis network model when the loss value stably reaches a certain expected minimum value;
the normalization processing is specifically to normalize the input body condition parameters and the corresponding body weight respectively, wherein the normalization methods are the same, and the normalization processing is specifically as follows:
wherein x and x' respectively represent a one-dimensional vector formed by body condition parameters of a certain head large animal before normalization and a one-dimensional vector formed by body condition parameters of the head large animal after normalization, min is a one-dimensional vector formed by a theoretical minimum value of each body condition of the large animal, and max is a one-dimensional vector formed by a theoretical maximum value of each body condition of the large animal;
the step of inputting the normalized body condition parameters into the multi-layer radial basis network model to obtain output data specifically comprises the following steps:
calculating the ith input data of the kth layer and the center point of each neuron of the kth layerEuropean distance->
The data is expanded by power series:mapping from low dimension to high dimension in an expanded form of polynomial kernels to extract more feature information, where n k Nerves representing the kth layerNumber of elements, x k Input data representing the kth layer, +.>Representing the kth layer ith neuron center point;
the dimension of the output data matrix is adjusted by the last full-connection layer, and the estimated weight y of the current multi-layer radial basis network model is output ′ And performing inverse normalization: y=y' (Weight) max -Weight min )+Weight min Obtaining an estimated body Weight of the large animal, wherein Weight is max And Weight min Theoretical body weight maximum and minimum for the large animal;
the loss function is specifically:
the function is expressed by expansion, and the general formula of the obtained network is as follows:
wherein L is a loss value obtained by a loss function, Y i For pre-measured standard weight data of the ith large animal, y i Weight data of the ith large animal estimated for the multi-layer radial basis network, N is the data amount input into the multi-layer radial basis network model, N k Representing the number of neurons of the k-th layer network, w k Full connection layer weight, x, representing a k-layer network k Representing the input of a layer k network,representing the center point of the jth neuron of the kth layer;
and taking the normalized body condition parameter result of the large animal to be predicted as input data, and inputting the input data into a trained multilayer radial base network model, wherein the output of the multilayer radial base network model is the estimated weight value of the large animal to be predicted.
2. The method for automatically assessing the weight of a large animal based on a multi-layer radial basis network according to claim 1, wherein the body condition parameters include body length, body height, body width, abdominal circumference, waist circumference.
3. The method for automatically estimating the body weight of the large animal based on the multilayer radial basis network according to claim 1, wherein the parameters for optimizing the multilayer radial basis network model by using the loss function are specifically as follows:
firstly, aiming at a loss function, taking a central point c of a radial base network as a variable to conduct gradient derivation, and subtracting a value obtained by derivation from the central point before optimization to realize updating of central point parameters;
then, aiming at the loss function, taking the radial basis neuron control number sigma as a variable to conduct gradient derivation, and subtracting the value obtained by the derivation from the sigma to conduct parameter updating;
finally, aiming at the loss function, carrying out gradient derivation by taking the full-connection layer neuron weight w as a variable, and subtracting a value obtained by derivation from the full-connection layer neuron weight to carry out parameter updating;
and continuously inputting a plurality of groups of large animal normalized body condition parameters and body weight for iteration for multiple times, and updating the central point, the neuron control number and the weight of the full-connection layer until a nonlinear model of stable expression body condition and body weight is obtained.
4. A large animal weight automatic assessment system based on a multilayer radial basis network, which is characterized by being applied to the large animal weight automatic assessment method based on the multilayer radial basis network as claimed in any one of claims 1-3, and comprising a construction module, a training module and a prediction module;
the construction module is used for generating center points in a random position mode according to the number of the set center points of each radial base layer, generating corresponding radial basis functions based on the center points, further obtaining radial basis neurons to form the radial base layers, connecting the radial base layers with the full-connection layers to generate a radial basis network model, and repeating the steps to construct the multi-layer radial basis network model;
the training module is used for training the multi-layer radial base network model, firstly inputting the normalized body condition parameters into the multi-layer radial base network model to obtain output data, inputting the output data and the standard data as parameters of a loss function to calculate a loss value, then optimizing the parameters of the multi-layer radial base network model by using the loss function, and obtaining the trained multi-layer radial base network model when the loss value is stable and reaches a certain expected minimum value;
the prediction module is used for taking the normalized body condition parameters of the large animal to be predicted as input, inputting the normalized body condition parameters into the trained multilayer radial basis network model, calculating a normalized body weight prediction result, and then carrying out inverse normalization on the result, wherein the output of the multilayer radial basis network model is used for obtaining the estimated body weight value of the large animal to be predicted.
5. A computer readable storage medium storing a program, wherein the program, when executed by a processor, implements the method for automatically assessing body weight of a large animal based on a multi-layered radial basis network according to any one of claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110854716.8A CN113487107B (en) | 2021-07-28 | 2021-07-28 | Automatic large animal weight assessment method, system and medium based on multilayer radial basis network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110854716.8A CN113487107B (en) | 2021-07-28 | 2021-07-28 | Automatic large animal weight assessment method, system and medium based on multilayer radial basis network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113487107A CN113487107A (en) | 2021-10-08 |
| CN113487107B true CN113487107B (en) | 2024-04-12 |
Family
ID=77943029
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110854716.8A Active CN113487107B (en) | 2021-07-28 | 2021-07-28 | Automatic large animal weight assessment method, system and medium based on multilayer radial basis network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113487107B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108416483A (en) * | 2018-04-27 | 2018-08-17 | 无锡南洋职业技术学院 | RBF type teaching quality evaluation prediction techniques based on PSO optimizations |
| CN109002942A (en) * | 2018-09-28 | 2018-12-14 | 河南理工大学 | A kind of short-term load forecasting method based on stochastic neural net |
| CN109144099A (en) * | 2018-08-28 | 2019-01-04 | 北京航空航天大学 | Unmanned aerial vehicle group action scheme fast evaluation method based on convolutional neural networks |
| CN110426112A (en) * | 2019-07-04 | 2019-11-08 | 平安科技（深圳）有限公司 | Live pig weight measuring method and device |
| CN110443364A (en) * | 2019-06-21 | 2019-11-12 | 深圳大学 | A kind of deep neural network multitask hyperparameter optimization method and device |
| CN110472280A (en) * | 2019-07-10 | 2019-11-19 | 广东工业大学 | A kind of power amplifier behavior modeling method based on generation confrontation neural network |
| CN111353631A (en) * | 2019-11-26 | 2020-06-30 | 国网山东省电力公司电力科学研究院 | Thermal power plant condenser vacuum degree prediction method based on multilayer LSTM |
| CN112037011A (en) * | 2020-08-14 | 2020-12-04 | 百维金科(上海)信息科技有限公司 | Credit scoring method based on FOA-RBF neural network |
| WO2021082809A1 (en) * | 2019-10-29 | 2021-05-06 | 山东科技大学 | Training optimization method for foreign exchange time series prediction |
| CN112862757A (en) * | 2021-01-14 | 2021-05-28 | 四川大学 | Weight evaluation system based on computer vision technology and implementation method |
| CN112967238A (en) * | 2021-02-23 | 2021-06-15 | 广东工业大学 | Porous medium permeability prediction method, electronic device and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190114531A1 (en) * | 2017-10-13 | 2019-04-18 | Cambia Health Solutions, Inc. | Differential equations network |
-
2021
- 2021-07-28 CN CN202110854716.8A patent/CN113487107B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108416483A (en) * | 2018-04-27 | 2018-08-17 | 无锡南洋职业技术学院 | RBF type teaching quality evaluation prediction techniques based on PSO optimizations |
| CN109144099A (en) * | 2018-08-28 | 2019-01-04 | 北京航空航天大学 | Unmanned aerial vehicle group action scheme fast evaluation method based on convolutional neural networks |
| CN109002942A (en) * | 2018-09-28 | 2018-12-14 | 河南理工大学 | A kind of short-term load forecasting method based on stochastic neural net |
| CN110443364A (en) * | 2019-06-21 | 2019-11-12 | 深圳大学 | A kind of deep neural network multitask hyperparameter optimization method and device |
| CN110426112A (en) * | 2019-07-04 | 2019-11-08 | 平安科技（深圳）有限公司 | Live pig weight measuring method and device |
| CN110472280A (en) * | 2019-07-10 | 2019-11-19 | 广东工业大学 | A kind of power amplifier behavior modeling method based on generation confrontation neural network |
| WO2021082809A1 (en) * | 2019-10-29 | 2021-05-06 | 山东科技大学 | Training optimization method for foreign exchange time series prediction |
| CN111353631A (en) * | 2019-11-26 | 2020-06-30 | 国网山东省电力公司电力科学研究院 | Thermal power plant condenser vacuum degree prediction method based on multilayer LSTM |
| CN112037011A (en) * | 2020-08-14 | 2020-12-04 | 百维金科(上海)信息科技有限公司 | Credit scoring method based on FOA-RBF neural network |
| CN112862757A (en) * | 2021-01-14 | 2021-05-28 | 四川大学 | Weight evaluation system based on computer vision technology and implementation method |
| CN112967238A (en) * | 2021-02-23 | 2021-06-15 | 广东工业大学 | Porous medium permeability prediction method, electronic device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113487107A (en) | 2021-10-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021098472A1 (en) | Crop yield estimation method based on deep temporal and spatial feature combined learning | |
| US20200342301A1 (en) | Convolutional neural network on-chip learning system based on non-volatile memory | |
| Grzesiak et al. | Examples of the use of data mining methods in animal breeding | |
| CN105105743A (en) | Intelligent electrocardiogram diagnosis method based on deep neural network | |
| CN107945182A (en) | Maize leaf disease recognition method based on convolutional neural networks model GoogleNet | |
| CN113449864A (en) | Feedback type pulse neural network model training method for image data classification | |
| Surya et al. | Cassava leaf disease detection using convolutional neural networks | |
| Chugh et al. | Potato leaf disease detection using inception V3 | |
| CN116960989B (en) | Power load prediction method, device and equipment for power station and storage medium | |
| CN109558889B (en) | Live pig comfort degree analysis method and device | |
| Brenner et al. | Tractable dendritic RNNs for reconstructing nonlinear dynamical systems | |
| CN116861967A (en) | Data enhancement method and device based on impulse neural network | |
| CN111047073B (en) | Aquaculture water quality prediction method and system based on neural network | |
| CN113487107B (en) | Automatic large animal weight assessment method, system and medium based on multilayer radial basis network | |
| Xiao et al. | Leaf disease detection based on lightweight deep residual network and attention mechanism | |
| CN116994144A (en) | High-resolution winter wheat yield estimation method, system, storage medium and electronic equipment based on synthetic aperture radar image | |
| CN115860113B (en) | Training method and related device for self-countermeasure neural network model | |
| CN109919005A (en) | Livestock personal identification method, electronic device and readable storage medium storing program for executing | |
| CN112149351B (en) | Microwave circuit physical dimension estimation method based on deep learning | |
| CN112906285B (en) | River and lake water system power reconstruction method, equipment and computer storage medium | |
| Usman et al. | Comparative study of artificial neural network algorithms performance for prediction of FL305DMY in crossbred cattle | |
| CN116703897B (en) | Pig weight estimation method based on image processing | |
| Lin et al. | Adaptive SVM-based classification systems based on the improved endocrine-based PSO algorithm | |
| Wang et al. | Vision-based measuring method for individual cow feed intake using depth images and a Siamese network | |
| Changji et al. | An image segmentation algorithm of corn disease based on the modified bionic pulse coupled neural network |
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 |