CN113807136A - Method, device, equipment and storage medium for detecting soil plowing state - Google Patents
Method, device, equipment and storage medium for detecting soil plowing state Download PDFInfo
- Publication number
- CN113807136A CN113807136A CN202010538117.0A CN202010538117A CN113807136A CN 113807136 A CN113807136 A CN 113807136A CN 202010538117 A CN202010538117 A CN 202010538117A CN 113807136 A CN113807136 A CN 113807136A
- Authority
- CN
- China
- Prior art keywords
- soil
- picture
- neural network
- network model
- block
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; 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 OR CALCULATING; 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Business, Economics & Management (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Agronomy & Crop Science (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Animal Husbandry (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method, a device, equipment and a storage medium for detecting the soil plowing state, wherein the method comprises the following steps: obtaining at least one soil picture in a target soil area; identifying at least one soil block included in the soil picture and the soil block size of each soil block; and determining the soil plowing state corresponding to the target soil area according to the number of the soil blocks of at least one size grade in the soil picture. The technical scheme of the embodiment of the invention solves the problem of low automation degree of the detection method of the soil plowing state in the prior art, can quickly display or inform the plowing state of the soil of a user, and improves the accuracy of plowing state detection.
Description
Technical Field
The embodiment of the invention relates to an image recognition and processing technology, in particular to a method, a device, equipment and a storage medium for detecting a soil plowing state.
Background
Turning is an important cultivation method in agricultural production, and mainly refers to shoveling, crushing and turning over soil. The plowing is beneficial to loosening soil, storing rainwater and promoting nutrient conversion and root system extension of crops; on the other hand, the plowing can effectively bury weeds, germs, pests and the like on the surface of soil into deep soil, improve the seeding quality of farmlands, and inhibit the growth and breeding of the diseases, the pests and the weeds. Therefore, the automatic and efficient detection of the soil plowing condition is very important by combining the development trend of the existing automatic agriculture.
The conventional method for detecting the plowing state is mainly observed by naked eyes of workers, has low automation degree, cannot rapidly display or inform the plowing condition, and has low detection accuracy.
Disclosure of Invention
The embodiment of the invention provides a method, a device, equipment and a storage medium for detecting a soil plowing state, which can rapidly display or inform a user of the plowing state of soil and improve the accuracy of detecting the plowing state.
In a first aspect, an embodiment of the present invention provides a method for detecting a soil plowing state, the method including:
obtaining at least one soil picture in a target soil area;
identifying at least one soil block included in the soil picture and the soil block size of each soil block;
and determining the soil plowing state corresponding to the target soil area according to the number of the soil blocks of at least one size grade in the soil picture.
Optionally, identifying at least one soil block included in the soil picture and a soil block size of each soil block includes:
inputting the soil picture into a pre-trained soil block recognition model, and acquiring at least one soil block included in the soil picture output from the soil block recognition model and the soil block size of each soil block.
Optionally, before acquiring at least one soil picture in the target soil area, the method further includes:
acquiring a plurality of turned soil sample pictures, wherein the soil sample pictures are pre-marked with at least one soil block position marking coordinate and soil block size grade;
dividing the multiple soil sample pictures into a training data set and a testing data set;
and performing iterative training on a neural network model by using the training data set and the test data set to obtain the soil mass recognition model.
Optionally, before iteratively training the neural network model by using the training data set and the test data set, the method further includes:
and loading the pre-weighting matched with the soil picture in the COCO data set, and presetting the neural network model.
Optionally, the iteratively training a neural network model by using the training data set and the test data set to obtain the soil block identification model includes:
dividing the training data set into a plurality of training data packets, and sequentially acquiring one training data packet;
respectively inputting each soil sample picture in the training data packet into the neural network model, and training the neural network model;
the neural network model identifies position coordinates according to soil blocks in a model output result, and adjusts parameters of the neural network model by using a random gradient descent method according to a distance loss function between the position coordinates and position marking coordinates in an input soil sample picture;
respectively inputting each soil sample picture in the test data set to the trained neural network model, and calculating the identification accuracy;
and if the calculation and recognition accuracy does not meet the preset accuracy requirement, returning to execute the operation of sequentially acquiring one training data packet until the trained neural network model meets the accuracy requirement.
Optionally, the neural network model is a convolutional neural network model constructed based on a MobileNet model and a YOLOV3 algorithm.
Optionally, determining a soil plowing state corresponding to the target soil area according to the number of soil blocks of at least one size grade in the soil picture, including:
judging whether the quantity of soil blocks with preset size grades in the soil picture is larger than a preset threshold value or not;
if so, determining that the soil plowing state corresponding to the target soil area is unqualified;
and if not, determining that the soil plowing state corresponding to the target soil area is qualified.
Optionally, the method for detecting the soil plowing state is performed by a mobile vision device, and the soil block identification model is pre-deployed in the mobile vision device;
obtaining at least one soil picture in a target soil area, comprising:
and acquiring a soil video in the target soil area acquired by the mobile vision equipment, and acquiring at least one video frame of the soil video as the soil picture.
In a second aspect, an embodiment of the present invention further provides a device for detecting a soil plowing state, the device including:
the soil picture acquisition module is used for acquiring at least one soil picture in the target soil area;
the soil block identification module is used for identifying at least one soil block included in the soil picture and the soil block size of each soil block;
and the soil plowing state determining module is used for determining the soil plowing state corresponding to the target soil area according to the number of the soil blocks with at least one size grade in the soil picture.
In a third aspect, an embodiment of the present invention further provides a computing device, where the computing device includes:
one or more processors;
storage means for storing one or more programs;
when the one or more programs are executed by the one or more processors, the one or more processors implement a method for detecting a state of soil turning provided by any embodiment of the invention.
In a fourth aspect, the embodiments of the present invention further provide a computer-readable storage medium, where the storage medium stores a computer program, and the computer program, when executed by a processor, implements a method for detecting a soil turning state provided by any of the embodiments of the present invention.
According to the technical scheme of the embodiment of the invention, at least one soil picture in the target soil area is obtained, the soil blocks included in the soil picture and the soil block size of each soil block are identified, and then the soil plowing state corresponding to the target soil area is determined according to the soil block number of at least one size grade in the soil picture. The technical scheme of the embodiment of the invention solves the problem of low automation degree of the detection method of the soil plowing state in the prior art, can quickly display or inform the plowing state of the soil of a user, and improves the accuracy of plowing state detection.
Drawings
FIG. 1 is a flow chart of a method for detecting a plowing state of soil according to one embodiment of the present invention;
FIG. 2 is a flow chart of a soil plowing state detection method according to a second embodiment of the present invention;
FIG. 3 is a structural view of a soil turning state detecting apparatus according to a third embodiment of the present invention;
fig. 4 is a schematic structural diagram of a computing device in the fourth embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Example one
Fig. 1 is a flowchart of a soil plowing state detection method according to an embodiment of the present invention, which is applicable to a situation where a soil plowing state is detected according to an image of a farmland, and the method can be executed by a soil plowing state detection device, which can be implemented by software and/or hardware, and can be generally integrated in a mobile vision device, such as a camera or a mobile terminal, and the method specifically includes the following steps:
and step 110, acquiring at least one soil picture in the target soil area.
In this step, the target soil region is a field soil region in which the user is concerned, specifically, a region in which the tillage work has been completed and a detection of the soil tillage state is waited for. And the picture acquired based on the target soil area is the soil picture.
Optionally, in this embodiment, a soil video in the target soil area may be first obtained, and then at least one video frame of the soil video is obtained as a soil picture; alternatively, the present embodiment may take multiple pictures of the plowed land as the soil picture in the target soil area.
Specifically, the soil video can be acquired from a third-party platform or can be directly shot through mobile visual equipment, and the soil video comprises a video shot for ploughed land.
And 120, identifying at least one soil block included in the soil picture and the soil block size of each soil block.
In this embodiment, a plurality of soil block images matching a plurality of soil block sizes of the soil blocks may be preset, the same soil block size may correspond to one or more soil block images, and then the plurality of soil block images are respectively matched with the soil picture, and according to a matching result, at least one soil block included in the soil picture and the soil block size of each soil block are identified.
For example, if a soil block image a corresponding to soil block size a matches the local image in the soil picture within the area a, it is determined that a soil block of soil block size a is identified within the area a.
In this embodiment, the soil picture may be input into a soil block recognition model trained in advance, and at least one soil block included in the soil picture output from the soil block recognition model and the soil block size of each soil block may be obtained.
In this embodiment, the soil block identification model is configured to perform depth analysis on the soil picture, generate a feature map reflecting multidimensional features of the soil picture, identify at least one soil block included in the soil picture and a soil block size of each soil block according to the feature map, and output an identification result.
Wherein the soil block size may be: the scale value of large, medium, small or fine may be a specific value such as 0.8cm (diameter), 0.7cm, 1.5cm, etc., which is not limited in this embodiment.
If the method provided by the embodiment of the invention is executed by mobile vision equipment, the soil block identification model needs to be deployed in the mobile vision equipment in advance so as to identify the obtained soil picture. Optionally, the method for establishing the soil block identification model includes the following steps:
s101, obtaining a plurality of turned soil sample pictures, wherein at least one position marking coordinate of soil blocks and the size grade of the soil blocks are marked in the soil sample pictures in advance;
in this step, the ploughed soil sample picture can be obtained from a third party platform or can be obtained through a shooting device, and the ploughed soil sample picture can be a soil picture in any farmland soil area.
The soil sample picture is pre-marked with position marking coordinates (typically, coordinates of four vertexes of a minimum circumscribed rectangle of the soil block) and soil block size grades (for example, large soil blocks, medium soil blocks, small soil blocks or small soil blocks).
Specifically, in this embodiment, the soil blocks in the soil sample picture are classified according to the soil block sizes in the soil sample picture, and the soil block size grades in the soil sample picture are labeled according to the soil block classification result.
S102, dividing the multiple soil sample pictures into a training data set and a testing data set;
in this step, in the present embodiment, a first number of soil sample pictures in the plurality of soil sample pictures are used as the training data set, and a second number of soil sample pictures are used as the testing data set. Typically, the first number is greater than the second number.
The training data set is used for being input into a pre-constructed neural network model, and parameters of the neural network model are adjusted according to an output result, so that training of the neural network model is completed;
and the test data set is used for testing the recognition accuracy of the trained neural network model, and determining whether the trained neural network model can be used as a soil mass recognition model according to the recognition accuracy.
S103, carrying out iterative training on the neural network model by using the training data set and the test data set to obtain the soil block identification model.
Typically, the neural Network model may be constructed based on a MobileNet model, a VGG (Visual Geometry Group Network) model, a Resnet model, or the like, and may perform target (soil block) detection using a YOLOV3 algorithm, an mtcnn (Multi-task convolutional neural Network) algorithm, or an SSD (Single Shot multiple box Detector) algorithm, which is not limited in this embodiment.
The neural network model comprises a data input layer, a hidden layer and an output layer, and is specifically used for realizing the maximum fitting between the output result of the output layer and the data of the input layer, and the hidden layer is used for updating the weight and parameters between the input layer and the output layer, so that the accuracy of the output result is ensured;
the YOLOV3 algorithm, the mtcnn algorithm, the SSD algorithm and the like are used for extracting deep features of the input data of the neural network model, and comparing the extracted features with features of a target to be identified, thereby identifying all targets in the input data.
The inventor finds out through multiple experiments that the recognition effect is optimal when the convolutional neural network model constructed based on the MobileNet model and the YOLOV3 algorithm is used as the neural network model, and correspondingly, in the step, the neural network model is the convolutional neural network model constructed based on the MobileNet model and the YOLOV3 algorithm.
In the embodiment, a pre-weight matched with the soil image in the COCO data set may be loaded, and the neural network model may be preset, so as to achieve an optimal model training effect with the least soil sample image.
The COCO data set is a data set comprising rich data sample pictures and training results obtained after training of the neural network model, and the COCO data set records the weight value of the neural network model corresponding to the best training result under various data samples.
After the neural network model is preset, the neural network model is iteratively trained by using the training data set and the test data set to obtain the soil block identification model.
In a specific embodiment, the training data set may be divided into a plurality of groups, a group of training data sets is first input into the neural network model, the neural network model identifies soil block parameters, such as the size and position coordinates of the soil block, of each soil block in the training data set according to the YOLOV3 algorithm, the identification result of the neural network model is compared with the soil block parameters labeled in the training data set, the difference between the recognition result of the neural network model and the actually marked soil block parameters is reduced by adjusting the parameters of the neural network model, then, the neural network model after parameter adjustment is used for identifying the test data set, whether the difference between the identification result and the soil block parameters actually marked in the test data set meets the preset requirement is judged, if so, finishing the training of the neural network model, and taking the neural network model as a soil mass recognition model; and if not, a new group of training data sets are taken to be input into the neural network model again, and the training process of the training data sets on the neural network model is continuously executed until the output of the trained neural network model aiming at the test data set meets the preset requirement.
After the soil block identification model is established, the present embodiment identifies all the soil blocks included in the soil picture and the soil block size of each soil block by using the soil block identification model.
And step 130, determining a soil plowing state corresponding to the target soil area according to the number of the soil blocks with at least one size grade in the soil picture.
In this embodiment, since the soil turning is a farming method of shoveling, crushing and turning the soil, when the turned soil block is too large, it indicates that the soil turning state is not good. The embodiment provides a method for determining a plowing state with soil according to the size and the number of soil blocks in a soil picture, and particularly, all the soil blocks included in the soil picture and the size of the soil blocks are identified in step 120, the embodiment classifies the soil blocks in the soil picture according to the size of the soil blocks in the soil picture, determines the size grade of each soil block in the soil picture according to the classification result of the soil blocks, and then judges whether the number of the soil blocks with at least one preset size grade in the soil picture is greater than a preset threshold value or not; if so, determining that the soil plowing state corresponding to the target soil area is unqualified; and if not, determining that the soil plowing state corresponding to the target soil area is qualified.
Specifically, if the soil mass size of the soil mass identified in the soil picture is directly the size grade, the number of soil masses of the preset size grade (for example, only the large soil mass or the large soil mass and the medium soil mass) may be counted, and if the soil mass size of the soil mass identified in the soil picture is a specific numerical value (for example, the soil mass diameter), the soil mass size in the form of each soil mass size grade may be determined first according to the numerical value range corresponding to the different size grades, for example, the large soil mass numerical value range is (1cm, 2cm), the medium soil mass data range is (0.8cm, 1cm), and the like, and then the number of soil masses of the preset size grade (for example, only the large soil mass or the large soil mass and the medium soil mass) may be counted.
According to the technical scheme of the embodiment of the invention, all the soil blocks and the sizes of the soil blocks in the soil picture are identified by obtaining the soil picture in the target soil area, and then the soil plowing state corresponding to the target soil area is determined according to the number of the soil blocks with at least one size grade. The technical scheme of the embodiment of the invention solves the problem of low automation degree of the detection method of the soil plowing state in the prior art, can quickly display or inform the plowing state of the soil of a user, and improves the accuracy of plowing state detection.
Example two
On the basis of the first embodiment, the embodiment provides a specific implementation mode for performing iterative training on the neural network model by using the training data set and the test data set to obtain the soil block identification model. The same or corresponding terms as those of the above-described embodiments are explained, and the description of the present embodiment is omitted. Fig. 2 is a flowchart of a method for detecting a soil plowing state according to a second embodiment of the present invention, as shown in fig. 2, the method includes the following steps:
In this step, the soil block identification model performs soil block detection on the soil picture according to the YOLOV3 algorithm, identifies at least one soil block included in the soil picture and the soil block size of each soil block, and outputs the identification result. Specifically, the method for establishing the soil block identification model comprises the following steps:
s201, obtaining a plurality of turned soil sample pictures, wherein at least one position marking coordinate of soil blocks and the size grade of the soil blocks are marked in the soil sample pictures in advance;
s202, dividing the multiple soil sample pictures into a training data set and a testing data set;
s203, dividing the training data set into a plurality of training data groups;
s204, sequentially acquiring one training data packet;
s205, respectively inputting each soil sample picture in the training data packet into the neural network model, and training the neural network model;
in this step, the present embodiment inputs each soil sample picture in the training data packet to the neural network model, and the neural network model identifies soil block parameters, such as the size and position coordinates of the soil block, of each soil block in the training data set according to the YOLOV3 algorithm, and outputs the soil block parameters of each soil block as a model output result.
After the neural network model obtains the soil block identification position coordinates in the model output result, calculating the soil block identification position coordinates in the output result, and a distance loss function between the soil block identification position coordinates and the position marking coordinates in the input soil sample picture, and then adjusting the parameters of the neural network model by using a random gradient descent method so as to enable the distance loss function to meet the preset requirement. And the position marking coordinates in the input soil sample picture are real soil block position marking coordinates marked in advance.
S206, respectively inputting each soil sample picture in the test data set into the trained neural network model, and calculating the identification accuracy.
S207, judging whether the calculation identification accuracy meets a preset accuracy requirement: if so, determining that the training of the neural network model is finished; otherwise, return to execute S204.
In a specific embodiment, after a training data packet trains a neural network model, this embodiment may input each soil sample picture in a test data set to the trained neural network model, calculate a distance loss function between a soil block identification position coordinate in an output result and a position marking coordinate in each input soil sample picture, calculate a ratio of the number of soil sample pictures satisfying a preset requirement to the number of all soil sample pictures in the test data set by the distance loss function, and use the ratio as the identification accuracy corresponding to the test data set. If the identification accuracy meets the preset accuracy requirement, taking the trained neural network model as a soil block identification model; and if the identification accuracy does not meet the preset accuracy requirement, returning to execute the operation of sequentially acquiring one training data packet until the trained neural network model meets the accuracy requirement.
After the soil block identification model is established, the present embodiment identifies all the soil blocks included in the soil picture and the soil block size of each soil block by using the soil block identification model.
And step 230, determining a soil plowing state corresponding to the target soil area according to the number of the soil blocks with at least one size grade in the soil picture.
According to the technical scheme of the embodiment of the invention, the soil picture in the target soil area is obtained, the soil picture is input into a pre-trained soil block identification model, all soil blocks included in the soil picture are identified in the soil block identification model, the soil block size of each soil block is utilized, and the soil plowing state corresponding to the target soil area is determined according to the soil block size and the number in the soil picture. The embodiment of the invention solves the problem of low automation degree of the detection method of the soil plowing state in the prior art, can quickly display or inform the plowing state of the soil of a user, and improves the accuracy of detecting the plowing state.
EXAMPLE III
Fig. 3 is a structural diagram of a soil plowing state detection device according to a third embodiment of the present invention, the device including: a soil picture acquisition module 310, a soil block identification module 320 and a soil plowing state determination module 330.
The soil picture acquiring module 310 is configured to acquire at least one soil picture in a target soil area; a soil block identification module 320, configured to identify at least one soil block included in the soil picture and a soil block size of each soil block; the soil plowing state determining module 330 is configured to determine a soil plowing state corresponding to the target soil area according to the number of soil blocks of at least one size grade in the soil picture.
According to the technical scheme of the embodiment of the invention, all the soil blocks and the sizes of the soil blocks in the soil picture are identified by obtaining the soil picture in the target soil area, and then the soil plowing state corresponding to the target soil area is determined according to the number of the soil blocks with at least one size grade. The technical scheme of the embodiment of the invention solves the problem of low automation degree of the detection method of the soil plowing state in the prior art, can quickly display or inform the plowing state of the soil of a user, and improves the accuracy of plowing state detection.
On the basis of the foregoing embodiments, the soil block identification module 320 may include:
and the soil picture input unit is used for inputting the soil picture into a pre-trained soil block identification model and acquiring at least one soil block included in the soil picture output from the soil block identification model and the soil block size of each soil block.
The soil picture acquiring module 310 may include:
the soil sample picture acquiring unit is used for acquiring a plurality of turned soil sample pictures, wherein the soil sample pictures are marked with at least one soil block position marking coordinate and soil block size grade in advance;
the soil sample picture dividing unit is used for dividing the multiple soil sample pictures into a training data set and a testing data set;
the neural network model training unit is used for carrying out iterative training on the neural network model by using the training data set and the test data set to obtain the soil mass recognition model;
the neural network model presetting unit is used for loading a pre-weighting value matched with the soil picture in the COCO data set and presetting the neural network model;
the training data set dividing unit is used for dividing the training data set into a plurality of training data packets and sequentially acquiring one training data packet;
the training data packet input unit is used for respectively inputting each soil sample picture in the training data packet into the neural network model and training the neural network model;
the neural network model identifies position coordinates according to soil blocks in a model output result, and adjusts parameters of the neural network model by using a random gradient descent method according to a distance loss function between the position coordinates and position marking coordinates in an input soil sample picture;
the recognition accuracy calculation unit is used for respectively inputting each soil sample picture in the test data set to the trained neural network model, calculating recognition accuracy, and if the calculation recognition accuracy does not meet the preset accuracy requirement, returning to execute the operation of sequentially acquiring one training data packet until the trained neural network model meets the accuracy requirement;
the neural network model building unit is used for building a convolutional neural network model based on a MobileNet model and a YOLOV3 algorithm;
and the soil video acquisition unit is used for acquiring a soil video in the target soil area acquired by the mobile visual equipment and acquiring at least one video frame of the soil video as a soil picture.
The soil plowing state determining module 330 may include:
and the judging unit is used for judging whether the quantity of the soil blocks with the preset size grade in the soil picture is larger than a preset threshold value, if so, determining that the soil plowing state corresponding to the target soil area is unqualified, and if not, determining that the soil plowing state corresponding to the target soil area is qualified.
The soil plowing state detection device provided by the embodiment of the invention can execute the soil plowing state detection method provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the execution method.
Example four
Fig. 4 is a schematic structural diagram of a computing apparatus according to a fourth embodiment of the present invention, as shown in fig. 4, the computing apparatus includes a processor 410, a memory 420, an input device 430, and an output device 440; the number of processors 410 in the computing device may be one or more, and one processor 410 is taken as an example in fig. 4; the processor 410, memory 420, input device 430, and output device 440 in the computing device may be connected by a bus or other means, such as by a bus in fig. 4.
The memory 420 serves as a computer-readable storage medium for storing software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to a method for detecting a soil turning state according to an embodiment of the present invention (for example, the soil picture acquisition module 310, the soil block identification module 320, and the soil turning state determination module 330 in a device for detecting a soil turning state). The processor 410 executes various functional applications and data processing of the computing device by executing software programs, instructions and modules stored in the memory 420, namely, implementing one of the soil plowing state detection methods described above. That is, the program when executed by the processor implements:
obtaining at least one soil picture in a target soil area;
identifying at least one soil block included in the soil picture and the soil block size of each soil block;
and determining the soil plowing state corresponding to the target soil area according to the number of the soil blocks of at least one size grade in the soil picture.
The memory 420 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function; the storage data area may store data created according to the use of the terminal, and the like. Further, the memory 420 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, memory 420 may further include memory located remotely from processor 410, which may be connected to a computing device over a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input device 430 may be used to receive input numeric or character information and generate key signal inputs related to user settings and function control of the computing device, and may include a keyboard and a mouse, etc. The output device 440 may include a display device such as a display screen.
EXAMPLE five
Fifth, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method according to any embodiment of the present invention. Of course, the embodiments of the present invention provide a computer-readable storage medium, which can perform operations related to a method for detecting a soil turning state according to any of the embodiments of the present invention. That is, the program when executed by the processor implements:
obtaining at least one soil picture in a target soil area;
identifying at least one soil block included in the soil picture and the soil block size of each soil block;
and determining the soil plowing state corresponding to the target soil area according to the number of the soil blocks of at least one size grade in the soil picture.
From the above description of the embodiments, it is obvious for those skilled in the art that the present invention can be implemented by software and necessary general hardware, and certainly, can also be implemented by hardware, but the former is a better embodiment in many cases. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as a floppy disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a FLASH Memory (FLASH), a hard disk or an optical disk of a computer, and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device) to execute the methods according to the embodiments of the present invention.
It should be noted that, in the above embodiment of the device for detecting a soil turning state, the included units and modules are only divided according to the functional logic, but not limited to the above division, as long as the corresponding functions can be realized; in addition, specific names of the functional units are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present invention.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (11)
1. A method for detecting a soil plowing state is characterized by comprising the following steps:
obtaining at least one soil picture in a target soil area;
identifying at least one soil block included in the soil picture and the soil block size of each soil block;
and determining the soil plowing state corresponding to the target soil area according to the number of the soil blocks of at least one size grade in the soil picture.
2. The method of claim 1, wherein identifying at least one soil block included in the soil picture and a soil block size for each soil block comprises:
inputting the soil picture into a pre-trained soil block recognition model, and acquiring at least one soil block included in the soil picture output from the soil block recognition model and the soil block size of each soil block.
3. The method of claim 2, further comprising, prior to obtaining at least one picture of soil within the target soil region:
acquiring a plurality of turned soil sample pictures, wherein the soil sample pictures are pre-marked with at least one soil block position marking coordinate and soil block size grade;
dividing the multiple soil sample pictures into a training data set and a testing data set;
and performing iterative training on a neural network model by using the training data set and the test data set to obtain the soil mass recognition model.
4. The method of claim 3, further comprising, prior to iteratively training a neural network model using the training dataset and the test dataset:
and loading the pre-weighting matched with the soil picture in the COCO data set, and presetting the neural network model.
5. The method of claim 3, wherein iteratively training a neural network model using the training dataset and the test dataset to obtain the soil mass recognition model comprises:
dividing the training data set into a plurality of training data packets, and sequentially acquiring one training data packet;
respectively inputting each soil sample picture in the training data packet into the neural network model, and training the neural network model;
the neural network model identifies position coordinates according to soil blocks in a model output result, and adjusts parameters of the neural network model by using a random gradient descent method according to a distance loss function between the position coordinates and position marking coordinates in an input soil sample picture;
respectively inputting each soil sample picture in the test data set to the trained neural network model, and calculating the identification accuracy;
and if the calculation and recognition accuracy does not meet the preset accuracy requirement, returning to execute the operation of sequentially acquiring one training data packet until the trained neural network model meets the accuracy requirement.
6. The method of claim 3, wherein the neural network model is a convolutional neural network model constructed based on a MobileNet model and a YOLOV3 algorithm.
7. The method of claim 1, wherein determining a soil turn-down status corresponding to the target soil area based on the number of clods of at least one size class in the soil picture comprises:
judging whether the quantity of the soil blocks with at least one preset size grade in the soil picture is greater than a preset threshold value or not;
if so, determining that the soil plowing state corresponding to the target soil area is unqualified;
and if not, determining that the soil plowing state corresponding to the target soil area is qualified.
8. The method of claim 2, wherein the method is performed by a mobile vision device and the soil mass identification model is pre-deployed in the mobile vision device;
obtaining at least one soil picture in a target soil area, comprising:
and acquiring a soil video in the target soil area acquired by the mobile vision equipment, and acquiring at least one video frame of the soil video as the soil picture.
9. A detection device for soil plowing state is characterized by comprising:
the soil picture acquisition module is used for acquiring at least one soil picture in the target soil area;
the soil block identification module is used for identifying at least one soil block included in the soil picture and the soil block size of each soil block;
and the soil plowing state determining module is used for determining the soil plowing state corresponding to the target soil area according to the number of the soil blocks with at least one size grade in the soil picture.
10. A computing device, wherein the computing device comprises:
one or more processors;
storage means for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement the method of detecting a state of soil plowing according to any one of claims 1 to 8.
11. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out a method of detecting a state of soil turning according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010538117.0A CN113807136A (en) | 2020-06-12 | 2020-06-12 | Method, device, equipment and storage medium for detecting soil plowing state |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010538117.0A CN113807136A (en) | 2020-06-12 | 2020-06-12 | Method, device, equipment and storage medium for detecting soil plowing state |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113807136A true CN113807136A (en) | 2021-12-17 |
Family
ID=78944087
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010538117.0A Pending CN113807136A (en) | 2020-06-12 | 2020-06-12 | Method, device, equipment and storage medium for detecting soil plowing state |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113807136A (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108289407A (en) * | 2015-09-18 | 2018-07-17 | 精密种植有限责任公司 | The control of devices, systems, and methods and farming tool for monitoring edaphic condition during farming operates |
| US20200005474A1 (en) * | 2018-06-27 | 2020-01-02 | Cnh Industrial Canada, Ltd. | Detecting and measuring the size of clods and other soil features from imagery |
-
2020
- 2020-06-12 CN CN202010538117.0A patent/CN113807136A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108289407A (en) * | 2015-09-18 | 2018-07-17 | 精密种植有限责任公司 | The control of devices, systems, and methods and farming tool for monitoring edaphic condition during farming operates |
| US20200005474A1 (en) * | 2018-06-27 | 2020-01-02 | Cnh Industrial Canada, Ltd. | Detecting and measuring the size of clods and other soil features from imagery |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Peeters et al. | Getis–Ord’s hot-and cold-spot statistics as a basis for multivariate spatial clustering of orchard tree data | |
| CN111553240B (en) | Corn disease condition grading method and system and computer equipment | |
| Castillejo-González et al. | Evaluation of pixel-and object-based approaches for mapping wild oat (Avena sterilis) weed patches in wheat fields using QuickBird imagery for site-specific management | |
| Liang et al. | StomataScorer: a portable and high‐throughput leaf stomata trait scorer combined with deep learning and an improved CV model | |
| CN114419367B (en) | A method and system for high-precision crop mapping | |
| Ramesh et al. | Comparison and analysis of rice blast disease identification in greenhouse controlled environment and field environment using ML algorithms | |
| CN114202691B (en) | Rice identification method, device and equipment based on remote sensing image | |
| Cui et al. | Weed identification in soybean seedling stage based on UAV images and Faster R-CNN | |
| CN113408374B (en) | Yield estimation method, device, equipment and storage medium based on artificial intelligence | |
| CN114612896B (en) | Rice yield prediction method, device and equipment based on remote sensing image | |
| CN118262149A (en) | A method and system for crop disease and insect pest image recognition based on deep learning | |
| Li et al. | An effective farmer-centred mobile intelligence solution using lightweight deep learning for integrated wheat pest management | |
| Che et al. | The effect of growth stage and plant counting accuracy of maize inbred lines on LAI and biomass prediction | |
| CN119624395A (en) | A full-process farmland management method and system based on deep learning | |
| CN113222017A (en) | Ground penetrating radar data generation method, control device, system and storage medium | |
| CN115690590B (en) | Crop growth abnormity monitoring method, device, equipment and storage medium | |
| Sodhi et al. | Robust plant phenotyping via model-based optimization | |
| Sun et al. | Counting wheat heads using a simulation model | |
| Li et al. | Disease recognition of maize leaf based on KNN and feature extraction | |
| Masykur et al. | Approach and analysis of yolov4 algorithm for rice diseases detection at different drone image acquisition distances | |
| CN119169473B (en) | Farmland soil exposure monitoring method and system based on time-series optical remote sensing images | |
| CN113807136A (en) | Method, device, equipment and storage medium for detecting soil plowing state | |
| CN119377427B (en) | Digital twinning-based GIS data visual management method and system | |
| CN115700805A (en) | Plant height detection method, device, equipment and storage medium | |
| CN109472771A (en) | Detection method, device and the detection device of maize male ears |
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 |