CN115393645A - Automatic soil classification and naming method and system, storage medium and intelligent terminal - Google Patents
Automatic soil classification and naming method and system, storage medium and intelligent terminal Download PDFInfo
- Publication number
- CN115393645A CN115393645A CN202211036173.XA CN202211036173A CN115393645A CN 115393645 A CN115393645 A CN 115393645A CN 202211036173 A CN202211036173 A CN 202211036173A CN 115393645 A CN115393645 A CN 115393645A
- Authority
- CN
- China
- Prior art keywords
- information
- value
- soil
- preset
- action
- 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
- 239000002689 soil Substances 0.000 title claims abstract description 114
- 238000000034 method Methods 0.000 title claims abstract description 47
- 230000009471 action Effects 0.000 claims description 84
- 230000006870 function Effects 0.000 claims description 47
- 238000012549 training Methods 0.000 claims description 42
- 239000011159 matrix material Substances 0.000 claims description 35
- 238000012545 processing Methods 0.000 claims description 21
- 238000004458 analytical method Methods 0.000 claims description 10
- 238000004088 simulation Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 5
- 230000000694 effects Effects 0.000 abstract description 3
- 238000005516 engineering process Methods 0.000 abstract description 2
- 230000000875 corresponding effect Effects 0.000 description 73
- 239000011435 rock Substances 0.000 description 15
- 230000008569 process Effects 0.000 description 8
- 239000003086 colorant Substances 0.000 description 2
- 239000003415 peat Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000004927 clay Substances 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/24—Earth materials
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Chemical & Material Sciences (AREA)
- Data Mining & Analysis (AREA)
- Remote Sensing (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Environmental & Geological Engineering (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geology (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Image Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to a method, a system, a storage medium and an intelligent terminal for automatically classifying and naming soil, which relate to the field of soil identification technology and comprise the steps of acquiring input image information; and inputting the input image information into a preset identification server for identification so as to output soil category information and soil parameter information. This application has the effect that improves soil nomination's overall efficiency.
Description
Technical Field
The application relates to the field of soil identification technology, in particular to a method, a system, a storage medium and an intelligent terminal for automatically classifying and naming soil.
Background
The soil components are complex, the types are numerous, and the classification emphasis points of different requirements are different, so that the soil is difficult to classify by using a uniform standard. At present, the method at home and abroad is to formulate a more definite and refined soil classification naming standard according to the characteristics of the soil on the basis of formulating a basic soil general classification standard and in various industries and regions, wherein the soil classification naming standard aims to better know and analyze the physical and chemical properties of the soil so as to meet the requirements of engineering application and scientific research.
In the related art, when the name of soil is assigned, the method generally adopts a mode of combining appearance checking and laboratory instrument component measurement: the appearance inspection is mainly performed by determining the type of the rock and soil through preliminary analysis on the granularity, the plasticity and the like of the rock and soil by professional technicians; the laboratory instrument analysis mainly measures the contents of organic matters, chemical components and the like of the rock and soil by means of a measuring instrument besides granularity, plasticity and the like, and classifies and names the rock and soil on the basis.
With respect to the related art described above, the inventors consider that the speed of naming soil by the above technique is slow, and there is room for improvement.
Disclosure of Invention
In order to improve the overall efficiency of soil naming, the application provides a method, a system, a storage medium and an intelligent terminal for automatic soil classification naming.
In a first aspect, the present application provides a method for automatically classifying and naming soil, which adopts the following technical scheme:
an automatic soil classification and naming method comprises the following steps:
acquiring input image information;
and inputting the input image information into a preset identification server for identification so as to output soil category information and soil parameter information.
By adopting the technical scheme, when the staff need to name the obtained rock sample, the image of the sample is input into the recognition server for rock recognition, and corresponding soil category information is output, so that the overall soil naming efficiency is high; meanwhile, corresponding soil parameters can be output, and workers can analyze whether the name is correct or not according to the soil parameters so as to artificially modify the incorrect condition and improve the name setting accuracy of the soil.
Optionally, the method for establishing the identification server includes:
acquiring requirement classification information, training sample image information and sample parameter information;
inputting training sample image information into a preset convolution module and a preset linear module to extract characteristic image information;
randomly generating each preset fixed state of the characteristic image information under the classification type corresponding to the requirement classification information and the action function value of the preset fixed action according to a preset random function, and defining the next fixed state of the current fixed state as a subsequent state;
determining a feedback value according to the fixed action in the current fixed state and the action under the classification corresponding to the requirement classification information in the sample parameter information, determining a maximum action function value in the fixed action in the subsequent state, and defining the action function value as a subsequent maximum value;
calculating according to a preset fixed value, a feedback value, an action function value and a subsequent maximum value to update the action function value of each fixed action, and performing matching analysis according to sample parameter information, fixed actions and simulated action values stored in a preset simulation database to determine the simulated action value of each fixed action under the sample parameter information;
calculating a difference value according to the action function value and the simulated action value to determine a loss value, and judging whether the loss values are all smaller than a preset qualified value;
if the loss values are all smaller than the qualified value, outputting a training completion signal and establishing an identification server according to the training completion signal;
and if the uneven loss value is smaller than the qualified value, correcting the weights in the convolution module and the linear module to re-determine the characteristic image information, and re-calculating the loss value until outputting a training completion signal.
By adopting the technical scheme, the network model training is carried out according to the training sample input by the staff, in the training process, the action function value under the actual condition is determined according to the corresponding action and the corresponding state, the accuracy condition of the network training is determined by comparing the actual action function value with the simulated action value, when the loss value is small, the rock classification and identification after the network training is more accurate, and at the moment, the identification server is established according to the condition to be used for rock soil identification.
Optionally, the method for extracting feature image information includes:
determining input matrix size information according to the sample image information;
inputting sample image information into a convolution module and calculating according to input matrix size information and preset convolution parameters to determine output matrix size information;
judging whether the value corresponding to the size information of the output matrix is smaller than a preset upper limit value or not;
if the value corresponding to the output matrix size information is smaller than the upper limit value, outputting a size satisfying signal;
if the numerical value corresponding to the output matrix size information is not smaller than the upper limit value, updating the output matrix size information into new input matrix size information to perform convolution calculation on the image again until the output size meets the signal;
after the size of the signal meets the requirement of signal output, matching analysis is carried out according to requirement classification information and model length information stored in a preset module database to determine model length information corresponding to the requirement classification information, and the sample image information processed by the convolution module is input into a linear model with the length corresponding to the model length information to output characteristic image information.
By adopting the technical scheme, the image can be subjected to feature extraction according to the image condition of the input sample, so that the image can be converted into the image which can be used for training the recognition server under the processing of the convolution module and the linear module.
Optionally, the method for updating the action function value includes:
matching and analyzing the demand classification information and the weight coefficient information stored in the preset weight database to determine the weight coefficient information corresponding to each demand classification information;
defining:
the value of the action function before update is q(s) t ，a t )；
The updated action function value is q * (s t ，a t )；
The subsequent maximum value is max(s) t+1 ，a n )；
The fixed value is alpha;
the feedback value is r;
the coefficient value corresponding to the weight coefficient information is gamma;
q * (s t ，a t )＝q(s t ，a t )+α[r+γ(max(s t+1 ，a n )-q(s t ，a t ))]。
by adopting the technical scheme, the corresponding weight value can be determined according to the classification condition of the rock and soil, and the updating calculation of the action function value can be realized according to the weight value.
In a second aspect, the present application provides an automatic soil classification and naming system, which adopts the following technical scheme:
an automatic soil classification and naming system, comprising:
the acquisition module is used for acquiring input image information;
the processing module is connected with the acquisition module and the judgment module and used for storing and processing information;
the judging module is connected with the acquiring module and the processing module and is used for judging the information;
the processing module inputs the input image information into a preset identification server for identification so as to output soil category information and soil parameter information.
By adopting the technical scheme, when the staff need to name the rock sample acquired by the acquisition module, the processing module inputs the image of the sample into the recognition server for rock recognition and outputs corresponding soil type information, so that the overall efficiency of soil naming is high; meanwhile, corresponding soil parameters can be output, and workers can analyze whether the name is correct or not according to the soil parameters so as to artificially modify the incorrect condition and improve the name setting accuracy of the soil.
In a third aspect, the present application provides an intelligent terminal, which adopts the following technical solution:
an intelligent terminal comprises a memory and a processor, wherein the memory stores a computer program which can be loaded by the processor and executes any one of the automatic soil classification and naming methods.
By adopting the technical scheme, through the use of the intelligent terminal, when a worker needs to name the obtained rock sample, the image of the sample is input into the recognition server for rock recognition, and corresponding soil category information is output, so that the overall efficiency of soil naming is high; meanwhile, corresponding soil parameters can be output, and workers can analyze whether the name is correct according to the soil parameters so as to manually modify the incorrect condition and improve the name fixing accuracy of the soil.
In a fourth aspect, the present application provides a computer storage medium, which can store a corresponding program, and has a feature of improving the overall efficiency of soil naming, and adopts the following technical scheme:
a computer readable storage medium storing a computer program capable of being loaded by a processor and executing any of the above-mentioned automatic soil classification and naming methods.
By adopting the technical scheme, the storage medium is provided with the computer program of the automatic soil classification and naming method, when a worker needs to name the obtained rock sample, the image of the sample is input into the recognition server to carry out rock recognition, and corresponding soil category information is output, so that the overall soil naming efficiency is high; meanwhile, corresponding soil parameters can be output, and workers can analyze whether the name is correct according to the soil parameters so as to manually modify the incorrect condition and improve the name fixing accuracy of the soil.
In summary, the present application includes at least one of the following beneficial technical effects:
1. the soil image needing to be named is input into the identification server, and the identification server can determine the soil name according to the self identification capability condition so as to realize the quick naming of the soil and improve the integral efficiency of the soil naming;
2. training the recognition server by using the training sample so as to improve the accuracy of the recognition server in naming the soil;
3. and the condition of the soil identification parameters is synchronously output when the soil is named, so that technicians can perform secondary judgment on the naming condition according to own experience, and when the naming is incorrect, the technicians can correct the naming condition in time.
Drawings
FIG. 1 is a flow chart of an automatic soil classification and naming method.
Fig. 2 is a flow chart of an identification server establishment method.
Fig. 3 is a flowchart of an image feature extraction method.
FIG. 4 is a block flow diagram of a method for automatic classification and naming of soil.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is further described in detail below with reference to fig. 1-4 and the embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The embodiments of the present invention will be described in further detail with reference to the drawings attached hereto.
The embodiment of the application discloses a method for automatically classifying and naming soil, wherein a recognition server is subjected to learning training according to training samples provided by working personnel, and when the recognition effect of the recognition server on each classification of soil is better, the recognition server is stated to be trained; in the process of identifying and naming soil, a worker only needs to input a soil image needing naming to the identification server, and the identification server inputs a corresponding soil name according to a training condition so as to realize quick naming of soil and improve the overall efficiency of naming soil.
Referring to fig. 1, the process of the method for automatically classifying and naming soil comprises the following steps:
step S100: input image information is acquired.
The image corresponding to the input image information is a soil image which is named by the worker, and the soil image can be shot and acquired through equipment with a shooting function on the worker.
Step S101: and inputting the input image information into a preset identification server for identification so as to output soil category information and soil parameter information.
The identification server is a server capable of identifying soil images and judging types, and can be established through learning training; the category corresponding to the soil category information is the name of soil, such as peat soil, silt clay and the like, and can be determined by analyzing 6 state categories such as color, strength, inclusion, plasticity, glossiness and uniformity of the soil; the parameters corresponding to the soil parameter information include specific parameters of the recognized soil image, for example, the type corresponding to the soil type information is peat soil, and at this time, if the black proportion is 0.8, red is 0.1, and gray is 0.1 in the parameters of the color state type, the soil is recognized as black, but the corresponding color proportion is output, so that an experienced worker can judge whether the soil name output in the recognition server is correct or not through analysis of the specific parameters of the state types such as color.
Referring to fig. 2, the method for establishing the recognition server includes:
step S200: and acquiring requirement classification information, training sample image information and sample parameter information.
The type corresponding to the requirement classification information is a state type required to be determined, and comprises 6 state types such as color, intensity, inclusion, plasticity, glossiness and uniformity of soil, the image corresponding to the training sample image information is an image which is input by a worker and used for training of the recognition server, the parameter corresponding to the sample parameter information is the specific situation of the training sample provided by the user under the type corresponding to the requirement classification information, for example, the color is gray, the intensity is good, the inclusion is a shell, the plasticity is plastic, the glossiness is rough and the uniformity is relatively uniform, and the sample image is synchronously input by the worker when being input.
Step S201: and inputting the training sample image information into a preset convolution module and a preset linear module to extract characteristic image information.
The convolution module is a module capable of performing convolution processing on the image, and the linear module is a module capable of fitting the image after the convolution processing into linearity, and are conventional modules in image processing, and are not described in detail; the matching use of the convolution module and the linear module is utilized to enable corresponding key features to be extracted from the training sample image, and the information of the feature image, namely the feature image information, is recorded.
Step S202: and randomly generating characteristic image information under each preset fixed state and the action function value of a preset fixed action under the classification type corresponding to the requirement classification information according to a preset random function, and defining the next fixed state of the current fixed state as a subsequent state.
The random function is a function capable of being assigned randomly, such as a greedy algorithm, the fixed state is an identification state, that is, a state of identifying an image corresponding to the current feature image information, the fixed action includes identification actions of 6 state categories such as color, intensity, inclusion, plasticity, glossiness and uniformity, taking color as an example, the fixed action includes all color actions, such as colors which may appear in soil such as gray, gray yellow, yellow and the like, and the action function value is a numerical value given to each fixed action under the random function in the fixed state; the current fixed state is a corresponding state of an image corresponding to the current feature image information, and a next fixed state of the current fixed state is a fixed state of an image corresponding to next feature image information of the image corresponding to the current feature image information.
Step S203: and determining a feedback value according to the fixed action in the current fixed state and the action under the classification corresponding to the requirement classification information in the sample parameter information, determining the maximum action function value in the fixed action in the subsequent state, and defining the action function value as the subsequent maximum value.
The action of the classification corresponding to the required classification information in the sample parameter information is the image type input by the user, for example, the classification is color, the corresponding action is gray, the color of the sample input by the user at the moment is gray, at the moment, if the fixed action in the current fixed state is consistent with gray, the identification server selects gray and selects gray correctly, at the moment, a correct corresponding feedback value is output, if the fixed action in the current fixed state is inconsistent with gray, the identification server selects gray and selects wrongly, at the moment, a feedback value corresponding to the mistake is output, the correct and wrong feedback values are determined by the staff according to the actual situation, repeated description is omitted, and the correct and wrong judgment can be determined through a training data set formulated by a technician; and a plurality of fixed actions exist in the subsequent state, and at the moment, the action function value with the maximum value is determined in all the fixed actions and is defined as the subsequent maximum value for identification, so that the data can be conveniently used in the subsequent process.
Step S204: and performing matching analysis according to the sample parameter information, the fixed actions and the simulated action values stored in the preset simulation database to determine the simulated action values of the fixed actions under the sample parameter information.
The fixed value is a fixed value set by a worker, the action function value can be updated and calculated according to the fixed value, the feedback value, the action function value and the subsequent maximum value so as to determine the accuracy of soil identification in the current network environment, and the calculation method comprises the following steps of: firstly, matching analysis is carried out according to the demand classification information and the weight coefficient information stored in a preset weight database to determine the weight coefficient information corresponding to each demand classification information, wherein the coefficient corresponding to the weight coefficient information is the weight value of one kind of the kind corresponding to the demand classification information in all kinds, and the influence degree of different kinds on the soil is determined through staff tests, for example, the color weight is 0.5, the strength weight is 0.1, and the inclusion weight is 0.04,The plasticity weight is 0.3, the glossiness weight is 0.01, the uniformity weight is 0.05, a weight database can be established through the corresponding relation, and the database establishing method is a conventional technical means of a person skilled in the art and is not described in detail; in this case, the value of the action function before update is defined as q(s) t ，a t ) (ii) a The updated action function value is q * (s t ，a t ) (ii) a The subsequent maximum value is max(s) t+1 ，a n ) (ii) a The fixed value is alpha; the feedback value is r; the coefficient value corresponding to the weight coefficient information is gamma; then q is * (s t ，a t )＝q(s t ，a t )+α[r+γ(max(s t+1 ，a n )-q(s t ，a t ))](ii) a The simulation action value is a perfect value which needs to be achieved by the network after the worker inputs the sample, for example, the color of the sample input by the worker is gray, the specified simulation action value is 1, that is, as long as the subsequent gray soil input is performed, the recognition accuracy can reach 100%, the corresponding relationship between the sample parameter information, the fixed action and the simulation action value is determined by the worker according to the test, the simulation database can be established through the corresponding relationship between the sample parameter information, the fixed action and the simulation action value, and the database establishment method is a conventional technical means of the worker in the field and is not described in detail.
Step S205: and calculating a difference value according to the action function value and the simulated action value to determine a loss value, and judging whether the loss values are all smaller than a preset qualified value.
The loss value is a difference value between the updated action function value and the simulated action value, namely, the inaccuracy rate of the updated action function value, the qualified value is the maximum value of the allowable loss value set by the worker, different types of the allowable loss value can correspond to different qualified values, and the allowable loss value is specifically set by the worker according to actual conditions; the purpose of the judgment is to know whether the current trained network meets the normal soil identification requirement.
Step S2051: if the loss values are all smaller than the qualified value, a training completion signal is output, and the recognition server is established according to the training completion signal.
When the loss values are all smaller than the qualified values, the identification of different types of soil states can be identified more correctly, a training completion signal is output at the moment to indicate that the network training is completed, and a corresponding identification server is established according to the condition so as to be used for daily identification and naming of the soil.
Step S2052: and if the uneven loss value is smaller than the qualified value, correcting the weights in the convolution module and the linear module to re-determine the characteristic image information, and re-calculating the loss value until outputting a training completion signal.
When the loss value is uneven and smaller than the qualified value, it is indicated that at least one variety in different soil types can not be identified more correctly, and at the moment, the weights in the convolution module and the linear module are corrected to change the output characteristic image, so that the action function value can be recalculated, the loss value can be updated, the image output condition is continuously updated, so that the network model can be continuously subjected to learning training until a training completion signal is output to determine a qualified network model; the method for correcting the weight is a routine operation for those skilled in the art, and is not described in detail.
Referring to fig. 3, the method for extracting feature image information includes:
step S300: input matrix size information is determined from the sample image information.
The matrix size corresponding to the input matrix size information is the size of the image picture corresponding to the sample image information, for example, when the input picture size is 224 × 3, the corresponding matrix size information corresponds to the size of 224.
Step S301: and inputting the sample image information into a convolution module, and calculating according to the input matrix size information and preset convolution parameters to determine output matrix size information.
The convolution parameter is each parameter value of convolution module, including convolution kernel size, stride and the number of zero-filling layers, and concrete numerical value is set for by the staff according to actual conditions, and the size that output matrix size information corresponds is the size of the image after the processing of convolution module, and the calculating method is: defining w as the size corresponding to the input matrix size information, w' as the size corresponding to the output matrix size information, k as the convolution kernel size, s as the stride, and p as the number of zero-padding layers, then
Step S302: and judging whether the value corresponding to the size information of the output matrix is smaller than a preset upper limit value or not.
The upper limit value is the maximum value of the matrix size of the image set by the operator when the feature extraction can be performed, and the purpose of the judgment is to know whether the image of the current feature extraction can be used for network training.
Step S3021: and if the value corresponding to the output matrix size information is smaller than the upper limit value, outputting a signal with the size meeting the requirement.
When the value corresponding to the output matrix size information is smaller than the upper limit value, the extracted image can be used for network training, and the output size meets the signal to identify the situation so as to facilitate subsequent processing.
Step S3022: and if the numerical value corresponding to the output matrix size information is not less than the upper limit value, updating the output matrix size information into new input matrix size information so as to perform convolution calculation on the image again until the output size meets the signal.
When the value corresponding to the output matrix size information is not smaller than the upper limit value, the image does not meet the requirement, the output matrix size information is updated to be new input matrix size information, the image is processed again in the convolution module until the output size meets the signal, and the image is used for network training.
Step S303: after the size of the signal meets the signal output requirement, matching analysis is carried out according to requirement classification information and model length information stored in a preset module database to determine model length information corresponding to the requirement classification information, and the sample image information processed by the convolution module is input into a linear model with the length corresponding to the model length information to output characteristic image information.
The length value corresponding to the simulation length information is a length value of a linear module, different requirement classification information corresponds to different simulation length information, for example, when the type is a color, the total number of the colors is 13, the corresponding length is 13, when the type is an intensity, the total number of the intensities is 3, the corresponding length is 3, the corresponding relationship between the two is set by a worker, and a corresponding module database is established through the relationship between the two; inputting the convolved image into a linear module with a corresponding length to output a corresponding characteristic image for network training of a type corresponding to different requirement classification information, and in the seventh step, converting the image of the convolved module into the linear module is a conventional technical means of a person skilled in the art and is not described in detail.
Referring to fig. 4, based on the same inventive concept, an embodiment of the present invention provides an automatic soil classification and naming system, including:
the acquisition module is used for acquiring input image information;
the processing module is connected with the acquisition module and the judgment module and used for storing and processing the information;
the judging module is connected with the acquiring module and the processing module and is used for judging the information;
the processing module inputs the input image information into a preset identification server for identification so as to output soil category information and soil parameter information;
the identification server establishing module is used for training the learning server according to the sample images input by the workers so that the server can accurately identify the soil images to be identified, and the identification server is established;
the image processing module is used for processing through the convolution module and the linear module according to the condition of an input image so as to enable the processed image to be used for network training;
and the action value calculation module is used for updating and calculating action function values of different actions so that the network model can be continuously trained according to the function condition.
It will be clear to those skilled in the art that, for convenience and simplicity of description, the foregoing division of the functional modules is merely used as an example, and in practical applications, the above function distribution may be performed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to perform all or part of the above described functions. For the specific working processes of the system, the apparatus and the unit described above, reference may be made to the corresponding processes in the foregoing method embodiments, and details are not described here again.
Embodiments of the present invention provide a computer-readable storage medium storing a computer program that can be loaded by a processor and execute a method for automatically classifying and naming soil.
Computer storage media include, for example: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
Based on the same inventive concept, the embodiment of the invention provides an intelligent terminal, which comprises a memory and a processor, wherein the memory is stored with a computer program which can be loaded by the processor and can execute the automatic soil classification and naming method.
It is obvious to those skilled in the art that, for convenience and simplicity of description, the above division of each functional module is only used for illustration, and in practical applications, the above function distribution may be performed by different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to perform all or part of the above described functions. For the specific working processes of the system, the apparatus and the unit described above, reference may be made to the corresponding processes in the foregoing method embodiments, and details are not described here again.
The foregoing is a preferred embodiment of the present application and is not intended to limit the scope of the present application in any way, and any features disclosed in this specification (including the abstract and drawings) may be replaced by alternative features serving equivalent or similar purposes, unless expressly stated otherwise. That is, unless expressly stated otherwise, each feature is only an example of a generic series of equivalent or similar features.
Claims (7)
1. A method for automatically classifying and naming soil is characterized by comprising the following steps:
acquiring input image information;
and inputting the input image information into a preset identification server for identification so as to output soil category information and soil parameter information.
2. The automatic soil classification and naming method according to claim 1, characterized in that the establishment method of the recognition server comprises the following steps:
acquiring requirement classification information, training sample image information and sample parameter information;
inputting training sample image information into a preset convolution module and a preset linear module to extract characteristic image information;
randomly generating each preset fixed state of the characteristic image information under the classification type corresponding to the requirement classification information and the action function value of the preset fixed action according to a preset random function, and defining the next fixed state of the current fixed state as a subsequent state;
determining a feedback value according to the fixed action in the current fixed state and the action under the classification corresponding to the requirement classification information in the sample parameter information, determining a maximum action function value in the fixed action in the subsequent state, and defining the action function value as a subsequent maximum value;
calculating according to a preset fixed value, a feedback value, an action function value and a subsequent maximum value to update the action function value of each fixed action, and performing matching analysis according to sample parameter information, fixed actions and simulated action values stored in a preset simulation database to determine the simulated action value of each fixed action under the sample parameter information;
calculating a difference value according to the action function value and the simulated action value to determine a loss value, and judging whether the loss values are all smaller than a preset qualified value;
if the loss values are all smaller than the qualified value, outputting a training completion signal and establishing an identification server according to the training completion signal;
if the loss value unevenness is smaller than the qualified value, the weights in the convolution module and the linear module are corrected to re-determine the characteristic image information, and the loss value calculation is carried out again until a training completion signal is output.
3. The automatic soil classification and naming method according to claim 2, characterized in that the extraction method of the feature image information comprises:
determining input matrix size information according to the sample image information;
inputting sample image information into a convolution module and calculating according to input matrix size information and preset convolution parameters to determine output matrix size information;
judging whether the value corresponding to the size information of the output matrix is smaller than a preset upper limit value or not;
if the value corresponding to the output matrix size information is smaller than the upper limit value, outputting a size satisfying signal;
if the value corresponding to the output matrix size information is not smaller than the upper limit value, updating the output matrix size information into new input matrix size information to perform convolution calculation on the image again until the output size meets the signal;
after the size of the signal meets the signal output requirement, matching analysis is carried out according to requirement classification information and model length information stored in a preset module database to determine model length information corresponding to the requirement classification information, and the sample image information processed by the convolution module is input into a linear model with the length corresponding to the model length information to output characteristic image information.
4. The automatic soil classification and naming method according to claim 3, characterized in that the updating method of the action function value comprises:
matching and analyzing the demand classification information and the weight coefficient information stored in the preset weight database to determine the weight coefficient information corresponding to each demand classification information;
defining:
the value of the action function before update is q(s) t ，a t )；
The updated action function value is q * (s t ，a t )；
The subsequent maximum value is max(s) t+1 ，a n )；
The fixed value is alpha;
the feedback value is r;
the coefficient value corresponding to the weight coefficient information is gamma;
q * (s t ，a t )＝q(s t ，a t )+α[r+γ(max(s t+1 ，a n )-q(s t ，a t ))]。
5. an automatic soil classification and naming system, comprising:
the acquisition module is used for acquiring input image information;
the processing module is connected with the acquisition module and the judgment module and used for storing and processing information;
the judging module is connected with the acquiring module and the processing module and is used for judging the information;
the processing module inputs the input image information into a preset identification server for identification so as to output soil category information and soil parameter information.
6. An intelligent terminal, comprising a memory and a processor, the memory having stored thereon a computer program that can be loaded by the processor and that executes the method according to any one of claims 1 to 4.
7. A computer-readable storage medium, in which a computer program is stored which can be loaded by a processor and which executes the method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211036173.XA CN115393645B (en) | 2022-08-27 | 2022-08-27 | Automatic soil classification and naming method, system, storage medium and intelligent terminal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211036173.XA CN115393645B (en) | 2022-08-27 | 2022-08-27 | Automatic soil classification and naming method, system, storage medium and intelligent terminal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115393645A true CN115393645A (en) | 2022-11-25 |
| CN115393645B CN115393645B (en) | 2024-08-13 |
Family
ID=84122613
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211036173.XA Active CN115393645B (en) | 2022-08-27 | 2022-08-27 | Automatic soil classification and naming method, system, storage medium and intelligent terminal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115393645B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116486140A (en) * | 2023-03-22 | 2023-07-25 | 中化现代农业有限公司 | Soil texture classification method and device and electronic equipment |
| CN117333486A (en) * | 2023-11-30 | 2024-01-02 | 清远欧派集成家居有限公司 | UV finish paint performance detection data analysis method, device and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107688856A (en) * | 2017-07-24 | 2018-02-13 | 清华大学 | Indoor Robot scene active identification method based on deeply study |
| CN110490100A (en) * | 2019-07-31 | 2019-11-22 | 中铁二院工程集团有限责任公司 | Ground automatic identification based on deep learning names method and system |
| US20200265308A1 (en) * | 2019-02-20 | 2020-08-20 | Fujitsu Limited | Model optimization method, data identification method and data identification device |
| CN111680698A (en) * | 2020-04-21 | 2020-09-18 | 北京三快在线科技有限公司 | Image recognition method and device and training method and device of image recognition model |
| CN113012771A (en) * | 2021-04-13 | 2021-06-22 | 广东工业大学 | Soil heavy metal spatial interpolation method and device and computer readable storage medium |
| WO2021208771A1 (en) * | 2020-04-18 | 2021-10-21 | 华为技术有限公司 | Reinforced learning method and device |
-
2022
- 2022-08-27 CN CN202211036173.XA patent/CN115393645B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107688856A (en) * | 2017-07-24 | 2018-02-13 | 清华大学 | Indoor Robot scene active identification method based on deeply study |
| US20200265308A1 (en) * | 2019-02-20 | 2020-08-20 | Fujitsu Limited | Model optimization method, data identification method and data identification device |
| CN110490100A (en) * | 2019-07-31 | 2019-11-22 | 中铁二院工程集团有限责任公司 | Ground automatic identification based on deep learning names method and system |
| WO2021208771A1 (en) * | 2020-04-18 | 2021-10-21 | 华为技术有限公司 | Reinforced learning method and device |
| CN111680698A (en) * | 2020-04-21 | 2020-09-18 | 北京三快在线科技有限公司 | Image recognition method and device and training method and device of image recognition model |
| CN113012771A (en) * | 2021-04-13 | 2021-06-22 | 广东工业大学 | Soil heavy metal spatial interpolation method and device and computer readable storage medium |
Non-Patent Citations (6)
| Title |
|---|
| CHENG-JIAN LIN: ""Using a Reinforcement Q-Learning-Based Deep Neural Network for Playing Video Games"", 《ELECTRONICS》, 7 October 2019 (2019-10-07), pages 1 - 15 * |
| LANJEWAR, M.G: ""Convolutional Neural Networks based classifications of soil images"", 《MULTIMEDIA TOOLS AND APPLICATIONS》, 14 February 2022 (2022-02-14), pages 10313 - 10336 * |
| 古 彭: ""Q-Learning算法的改进和实现"", 《计算机科学与应用》, 28 July 2021 (2021-07-28), pages 1 - 14 * |
| 吴克宁: ""土壤质地分类及其在我国应用探讨"", 《土壤学报》, 31 January 2019 (2019-01-31), pages 227 - 241 * |
| 郝慧珍: ""砂岩显微图像分析方法及其工具实现"", 《计算机科学》, 30 November 2017 (2017-11-30), pages 50 - 55 * |
| 雷林建;孙胜利;向玉开;张悦;刘会凯;: "智能制造中的计算机视觉应用瓶颈问题", 中国图象图形学报, no. 07, 16 July 2020 (2020-07-16), pages 52 - 65 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116486140A (en) * | 2023-03-22 | 2023-07-25 | 中化现代农业有限公司 | Soil texture classification method and device and electronic equipment |
| CN116486140B (en) * | 2023-03-22 | 2024-04-05 | 中化现代农业有限公司 | Soil texture classification method and device and electronic equipment |
| CN117333486A (en) * | 2023-11-30 | 2024-01-02 | 清远欧派集成家居有限公司 | UV finish paint performance detection data analysis method, device and storage medium |
| CN117333486B (en) * | 2023-11-30 | 2024-03-22 | 清远欧派集成家居有限公司 | UV finish paint performance detection data analysis method, device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115393645B (en) | 2024-08-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115393645A (en) | Automatic soil classification and naming method and system, storage medium and intelligent terminal | |
| CN103528617B (en) | A kind of cockpit instrument identifies and detection method and device automatically | |
| CN108960269B (en) | Feature acquisition method and device for data set and computing equipment | |
| CN110232445B (en) | Cultural relic authenticity identification method based on knowledge distillation | |
| CN109523188B (en) | Man-machine interface display-oriented method and system for evaluating man-machine cognitive characteristic work efficiency | |
| CN114519498B (en) | Quality inspection and evaluation method and system based on BIM model | |
| CN114414935A (en) | Automatic positioning method and system for feeder fault area of power distribution network based on big data | |
| CN112907026A (en) | Comprehensive evaluation method based on editable mesh index system | |
| CN111612149A (en) | Main network line state detection method, system and medium based on decision tree | |
| CN112527676A (en) | Model automation test method, device and storage medium | |
| CN111626358A (en) | Tunnel surrounding rock grading method based on BIM picture recognition | |
| CN117730655B (en) | Quantitative analysis method, device, equipment and storage medium for vigor of rice seeds | |
| CN111414930B (en) | Deep learning model training method and device, electronic equipment and storage medium | |
| CN112529112B (en) | Mineral identification method and device | |
| CN109946636A (en) | A kind of digital multimeter indicating value automatic calibrating method based on deep learning | |
| CN115994742B (en) | Full life cycle management method and device for wet-mixed mortar plasticizer | |
| CN116596395B (en) | Operation quality control platform for engineering project evaluation unit guidance and detection | |
| CN112614570A (en) | Sample set labeling method, pathological image classification method and classification model construction method and device | |
| CN116958053A (en) | Bamboo stick counting method based on yolov4-tiny | |
| CN114708445B (en) | Trademark similarity recognition method and device, electronic equipment and storage medium | |
| CN112505337B (en) | Data processing method for assisting analysis of sample | |
| CN114155412A (en) | Deep learning model iteration method, device, equipment and storage medium | |
| CN111598418A (en) | Balance-based item sorting method, balance-based item sorting device, balance-based item sorting equipment and storage medium | |
| CN111738410A (en) | Beef cattle individual growth curve acquisition method and device and storage medium | |
| CN115277487B (en) | Communication processing system of electronic terminal |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 19-2, No. 288 and 296, Baihe Road, Ningbo High tech Zone, Ningbo, Zhejiang 315,000 Applicant after: Ningbo East China Nuclear Industry Survey and Design Institute Group Co.,Ltd. Address before: 19-2, No. 288 and 296, Baihe Road, Ningbo High tech Zone, Ningbo, Zhejiang 315,000 Applicant before: Ningbo East China Nuclear Industry Engineering Survey Institute |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |