CN107742290A - Plant disease identifies method for early warning and device - Google Patents
Plant disease identifies method for early warning and device Download PDFInfo
- Publication number
- CN107742290A CN107742290A CN201710968700.3A CN201710968700A CN107742290A CN 107742290 A CN107742290 A CN 107742290A CN 201710968700 A CN201710968700 A CN 201710968700A CN 107742290 A CN107742290 A CN 107742290A
- Authority
- CN
- China
- Prior art keywords
- disease
- plant
- images
- recognized
- image
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30188—Vegetation; Agriculture
Abstract
The present invention provides a kind of plant disease identification method for early warning and device, is related to agricultural plant protection technical field.This method and device include corresponding spectrum textural characteristics by obtaining colored images to be recognized, the images to be recognized；Deep learning identification model after the input training of the images to be recognized of acquisition is identified classification, the deep learning identification model is according to the spectrum textural characteristics of the images to be recognized to obtain the severity of disease species and disease；Corresponding early warning is sent according to the severity.Scheme provided by the invention can reduce the manpower and material resources of identification plant disease, and improve the degree of accuracy and the recognition speed of identification plant disease, in addition, the program also can obtain the severity of plant disease and send corresponding early warning according to the severity, contribute to related personnel to carry out preventing and treating control to plant disease in time.
Description
Technical field
The present invention relates to agricultural plant protection technical field, and method for early warning and dress are identified in particular to a kind of plant disease
Put.
Background technology
Plant disease identifies and preventing and treating is an important step of agricultural production.For example, the pathogen of plant is classified
The maturity of corresponding pathogen is can obtain, the severity etc. of disease is can obtain according to plant severity Scaling.In the prior art,
It is generally necessary to which observing and predicting personnel carries out manual research in field, the different personnel that observe and predict differ greatly in investigation time stage standard, meanwhile,
Disease species are more, require high to Forecast Techniques personnel professional ability.In most cases, Forecast Techniques personnel need to combine books
Atlas, morphological feature description etc. are identified, and manual identified easily malfunctions, and recognition efficiency is low, and interference from human factor is big, no
The requirement for quickly and accurately identifying plant thing disease species and severity can be met.Therefore, how to provide one kind can solve it is above-mentioned
The method and device of problem, it has also become the technical problem of those skilled in the art's urgent need to resolve.
The content of the invention
In order to overcome above-mentioned deficiency of the prior art, the present invention provides a kind of plant disease identification method for early warning and dress
Put, plant disease can be identified and according to the severity grade of the disease of identification and send corresponding early warning, and then
Solve the above problems.
To achieve these goals, the technical scheme that present pre-ferred embodiments are provided is as follows：
For method, present pre-ferred embodiments provide a kind of plant disease identification method for early warning, applied to phytopathy
Evil identification early warning system, the plant disease identification early warning system include deep learning identification model, and methods described includes：
Colored images to be recognized is obtained, the images to be recognized includes corresponding spectrum-textural characteristics；
Classification, the depth is identified in deep learning identification model after the images to be recognized input training of acquisition
Degree learns identification model according to spectrum-textural characteristics of the images to be recognized to obtain the severity of disease species and disease；
Corresponding early warning is sent according to the severity.
In the preferred embodiment, the deep learning after the above-mentioned images to be recognized input training by acquisition
Identification model was identified before the step of classification, and methods described includes：
The training image collection of plant disease, including multiple training subgraphs are obtained, each training subgraph includes sense
Catch an illness the default labels of harmful plant image and corresponding disease species, wherein, the plant image includes the blade of plant
At least one of image, petal image, fruit image, limb/branch image, rhizome image, the default label include disease
Evil species, disease severity grade or pathogen development rank；
Using the training image collection, using deep learning identification model described in deep learning Algorithm for Training, trained
Deep learning identification model afterwards.
In the preferred embodiment, the deep learning after the above-mentioned images to be recognized input training by acquisition
Identification model is identified classification, the deep learning identification model according to spectrum-textural characteristics of the images to be recognized with
The step of obtaining the severity of disease species and disease, including：
Training image in spectrum-textural characteristics of the images to be recognized and the deep learning identification model is concentrated
The spectrum textural characteristics of training subgraph matched, it is similar to the training subgraph to obtain the images to be recognized
Degree；
Concentrated in the training image and choose the maximum training subgraph of similarity, the training subgraph that will be selected
Physical tags of the default label of picture as the disease of the images to be recognized, to obtain plant infection in the images to be recognized
Disease species, disease severity grade or pathogen development rank and introduced disease region.
In the preferred embodiment, the above method also includes：
At least one images to be recognized corresponding to one species preset number plant is associated；
For at least one images to be recognized corresponding to one species preset number plant, according to obtained disease kind
Class, disease severity grade or the region of pathogen development rank and introduced disease, calculate the disease of the species preset number plant
Feelings index or pathogen spore maturation index, with obtain the disease index of the species preset number plant or pathogen spore into
Ripe degree index.
In the preferred embodiment, above-mentioned plant disease identification early warning system includes server and the service
The user terminal of device communication connection and the image collecting device being connected with the server communication, the deep learning identification model
It is arranged on the server, the coloured image that described image harvester is used to gather the plant of introduced disease is using as described
Images to be recognized, and the images to be recognized is sent to the deep learning identification model in the server.
In the preferred embodiment, the above method also includes：
The disease species in the default label are associated with corresponding disease control strategy in advance；
Local current weather information is obtained, according to disease corresponding to the weather information and multiple images to be recognized
Severity calculates fashion trend of the disease in local preset time period；
According to the fashion trend, disease control strategy and/or early warning corresponding to the disease are sent to the user terminal
Prompting.
For device, presently preferred embodiments of the present invention provides a kind of plant disease identification prior-warning device, applied to plant
Disease recognition early warning system, the plant disease identification early warning system include deep learning identification model, and the plant disease is known
Other prior-warning device includes：
First acquisition unit, for obtaining the images to be recognized of colour, the images to be recognized includes corresponding spectrum-line
Manage feature；
Taxon is identified, is entered for the deep learning identification model after the images to be recognized input training by acquisition
Row identification classification, the deep learning identification model is according to spectrum-textural characteristics of the images to be recognized to obtain disease kind
The severity of class and disease；
Early warning unit, for sending corresponding early warning according to the severity.
In the preferred embodiment, above-mentioned plant disease identification prior-warning device also includes：
Second acquisition unit, it is each described for obtaining the training image collection of plant disease, including multiple training subgraphs
Subgraph is trained to include the plant image of introduced disease and the default label of corresponding disease species, wherein, the plant figure
As at least one of the leaf image including plant, petal image, fruit image, limb/branch image, rhizome image, institute
Stating default label includes disease species, disease severity grade or pathogen development rank；
Model training unit, for using the training image collection, using deep learning described in deep learning Algorithm for Training
Identification model, the deep learning identification model after being trained.
In the preferred embodiment, above-mentioned identification taxon is additionally operable to：
Training image in spectrum-textural characteristics of the images to be recognized and the deep learning identification model is concentrated
The spectrum textural characteristics of training subgraph matched, it is similar to the training subgraph to obtain the images to be recognized
Degree；
Concentrated in the training image and choose the maximum training subgraph of similarity, the training subgraph that will be selected
Physical tags of the default label of picture as the disease of the images to be recognized, to obtain plant infection in the images to be recognized
Disease species, disease severity grade or pathogen development rank and introduced disease region.
In the preferred embodiment, above-mentioned plant disease identification prior-warning device also includes：
Associative cell, at least one images to be recognized corresponding to one species preset number plant to be associated；
Computing unit, for at least one images to be recognized, root corresponding to one species preset number plant
According to obtained disease species, disease severity grade or the region of pathogen development rank and introduced disease, it is pre- to calculate the species
If the disease index or pathogen spore maturation index of number plant, to obtain the disease index of the species preset number plant
Or pathogen spore maturation index.
In terms of existing technologies, plant disease identification method for early warning and device provided by the invention at least have following
Beneficial effect：For methods described by obtaining colored images to be recognized, it is special that the images to be recognized includes corresponding spectrum-texture
Sign；Classification, the depth is identified in deep learning identification model after the images to be recognized input training of acquisition
Identification model is practised according to spectrum-textural characteristics of the images to be recognized to obtain the severity of disease species and disease；According to
The severity sends corresponding early warning.Scheme provided by the invention can reduce the manpower and material resources of identification plant disease,
And the degree of accuracy and the recognition speed of identification plant disease are improved, in addition, the program also can obtain the severity and root of plant disease
Early warning is sent according to the severity, contributes to related personnel to carry out preventing and treating control to plant disease in time.
To enable the above objects, features and advantages of the present invention to become apparent, present pre-ferred embodiments cited below particularly,
And accompanying drawing appended by coordinating, it is described in detail below.
Brief description of the drawings
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below by embodiment it is required use it is attached
Figure is briefly described.It should be appreciated that the following drawings illustrate only certain embodiments of the present invention, therefore it is not construed as pair
The restriction of scope, for those of ordinary skill in the art, on the premise of not paying creative work, can also be according to this
A little accompanying drawings obtain other related accompanying drawings.
Fig. 1 is the interaction schematic diagram that the plant disease that present pre-ferred embodiments provide identifies early warning system.
Fig. 2 is the block diagram for the server that present pre-ferred embodiments provide.
Fig. 3 is that the plant disease that present pre-ferred embodiments provide identifies one of schematic flow sheet of method for early warning.
Fig. 4 is the two of the schematic flow sheet that the plant disease that present pre-ferred embodiments provide identifies method for early warning.
Fig. 5 is the block diagram that the plant disease that present pre-ferred embodiments provide identifies prior-warning device.
Icon：10- plant diseases identify early warning system；100- user terminals；200- image collecting devices；300- is serviced
Device；310- processing units；320- communication units；330- memory cell；400- networks；500- plant diseases identify prior-warning device；
510- second acquisition units；520- model training units；530- first acquisition units；540- identifies taxon；550- early warning
Tip element；560- associative cells；570- computing units.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation describes.Obviously, described embodiment is only the part of the embodiment of the present invention, rather than whole embodiments.It is logical
The component for the embodiment of the present invention being often described and illustrated herein in the accompanying drawings can be configured to arrange and design with a variety of.
Therefore, below the detailed description of the embodiments of the invention to providing in the accompanying drawings be not intended to limit it is claimed
The scope of the present invention, but be merely representative of the present invention selected embodiment.Based on embodiments of the invention, people in the art
The every other embodiment that member is obtained on the premise of creative work is not made, belongs to the scope of protection of the invention.
It should be noted that：Similar label and letter represents similar terms in following accompanying drawing, therefore, once a certain Xiang Yi
It is defined, then it further need not be defined and explained in subsequent accompanying drawing in individual accompanying drawing.In addition, term " the
One ", " second " etc. is only used for distinguishing description, and it is not intended that instruction or hint relative importance.
Below in conjunction with the accompanying drawings, some embodiments of the present invention are elaborated.It is following in the case where not conflicting
Feature in embodiment and embodiment can be mutually combined.
Fig. 1 is refer to, is the interaction schematic diagram for the plant disease identification early warning system 10 that present pre-ferred embodiments provide.
In embodiments of the present invention, plant disease identification early warning system 10 can include server 300, image collecting device 200,
And user terminal 100.Server 300 can be communicated to connect with least one image collecting device 200 by network 400, be used for
Classification is identified in the images to be recognized sent to image collecting device 200, to obtain the species of plant disease and the disease
Corresponding severity, and corresponding early warning is sent according to severity.
Certainly, in other embodiments, server 300 can also obtain image collecting device indirectly by other means
200 images to be recognized gathered.For example, image collecting device 200 communicates to connect with user terminal 100, collection can be treated
Identification image is sent directly to user terminal 100, or the images to be recognized that image collecting device 200 gathers can be situated between by storing
Matter (such as USB flash disk) copy is stored in user terminal 100.The images to be recognized that namely image collecting device 200 gathers can be direct
Or be stored in user terminal 100 indirectly, then by user terminal 100 by the images to be recognized send to server 300 with
Disease is identified classification.
Wherein, the images to be recognized can be the colored plant image infected with disease.The plant image includes, but not
It is limited at least one of the leaf image of plant, petal image, fruit image, limb/branch image, rhizome image, to make
For investigation sample.
In the present embodiment, the user terminal 100 may be, but not limited to, smart mobile phone, PC
(personal computer, PC), tablet personal computer, personal digital assistant (personal digital assistant, PDA),
Mobile internet surfing equipment (mobile Internet device, MID) etc..The network 400 may be, but not limited to, cable network
Or wireless network.
In the present embodiment, described image harvester 200 can be general camera, high definition camera etc..In addition, the figure
As harvester 200 and user terminal 100 can be identical equipment (such as smart mobile phone), or different equipment.
For example, described image harvester 200 can be the combination of other assemblies.For example image collecting device 200 can be by taking the photograph
Picture head and the combined formation of microscope, the image of the pathogen of plant infection under microscope can be obtained according to demand, to take
Being engaged in device 300 can be according to the species and development rank of the pathogen of pathogen image recognition plant infection.Wherein, the pathogen
A variety of sickle-like bacteria (such as yellow fusarium, Fusorium moniliforme Sheldon etc.), handle rest fungus, smut etc. can be included but is not limited to, can be by aobvious
Images to be recognized corresponding to micro mirror acquisition, the development rank include incubation period, incubation period, the phase of originating, full incidence period and decline phase
Deng.
Fig. 2 is refer to, is the block diagram for the server 300 that present pre-ferred embodiments provide.In the present embodiment,
The server 300 may include processing unit 310, communication unit 320, memory cell 330 and plant disease identification early warning dress
Put 500.Wherein, processing unit 310, communication unit 320, memory cell 330 and plant disease identification prior-warning device 500 are each
Directly or indirectly it is electrically connected between element, to realize the transmission of data or interaction.For example, these elements can lead between each other
Cross one or more communication bus or signal wire is realized and is electrically connected with.
The processing unit 310 can be central processing unit (Central Processing Unit, CPU), network processes
Device (Network Processor, NP), graphics processor (Graphics Processing Unit, GPU) etc.；It can also be
Digital signal processor (DSP), application specific integrated circuit (ASIC), field programmable gate array (FPGA) or other programmable patrol
Collect device, discrete gate or transistor logic, discrete hardware components.It can realize or perform in the embodiment of the present invention
Disclosed each method, step and logic diagram.
The communication unit 320 is used for the communication link that image collecting device 200 and server 300 are established by network 400
Connect, and pass through the transceiving data of network 400.
The memory cell 330 may be, but not limited to, random access memory, read-only storage, may be programmed read-only deposit
Reservoir, Erasable Programmable Read Only Memory EPROM, Electrically Erasable Read Only Memory etc..In the present embodiment, the storage
Unit 330 can be used for the images to be recognized of the collection of storage image harvester 200.Certainly, the memory cell 330 can be with
For storage program, the processing unit 310 performs the program after execute instruction is received.
It is understood that the structure shown in Fig. 2 is only a kind of structural representation of server 300, the server 300 is also
It can include than more or less components shown in Fig. 2.Each component shown in Fig. 2 can use hardware, software or its group
Close and realize.
Refer to Fig. 3, be present pre-ferred embodiments provide plant disease identification method for early warning schematic flow sheet it
One.In the present embodiment, plant disease identification method for early warning can be applied to above-mentioned plant disease identification early warning system 10.Should
Plant disease identification early warning system 10 includes deep learning identification model, classification can be identified to above-mentioned images to be recognized, with
Severity corresponding to the species and the disease of the disease of plant infection is obtained, and then to sending corresponding early warning, so that phase
Pass personnel are prevented and treated plant disease in time.
Wherein, deep learning identification model may be provided at server 300 and/or user terminal 100.Alternatively, the depth
Study identification model is arranged on server 300, to improve the speed to disease recognition classification.The severity can be regarded as
The development rank of pathogen is stated, available for the developmental state that plant infection disease is identified for single images to be recognized.
The idiographic flow of method for early warning and step, which are described in detail, to be identified to the plant disease shown in Fig. 3 below.
In the present embodiment, the plant disease identification method for early warning may comprise steps of：
Step S630, obtains the images to be recognized of colour, and the images to be recognized includes corresponding spectrum-textural characteristics.
In the present embodiment, plant infection different diseases will cause external appearance characteristic that corresponding change occurs.For example, wheat holds
Water stain shape brown scab is presented in easy infection head blight, sick glume base portion at fringe initial stage, and the awn of wheat is withered, is gradually extended to whole small ear,
And other small ears can be spread to, pink mustiness thing is produced in the commissure of glume later, the later stage causes part small ear or full fringe withered
Extremely.That is, infect that the sick fringe severity of head blight is associated with the illness that wheat shows, can be by obtaining the coloured silk of wheat introduced disease
Color image, as images to be recognized, the spectrum textural characteristics based on the images to be recognized, it can obtain the disease kind that wheat is infected
Class and severity.Equally, the head blight shell of ascus of different development ranks, can be adjusted by the training and identification of characteristics of image
Look into ascospore mature condition of sample etc..
Step S640, the deep learning identification model after the input training of the images to be recognized of acquisition is identified point
Class, the deep learning identification model is according to spectrum-textural characteristics of the images to be recognized to obtain disease species and disease
Severity.
Refer to Fig. 4, be present pre-ferred embodiments provide plant disease identification method for early warning schematic flow sheet it
Two.In the present embodiment, before step S640, this method can include step S610 and step S620.Wherein, step
S610 and step S620 are before step S630.
Step S610, the training image collection of plant disease, including multiple training subgraphs are obtained, each training subgraph
As including the plant image of introduced disease and the default label of corresponding disease species, wherein, the plant image includes planting
At least one of the leaf image of thing, petal image, fruit image, limb/branch image, rhizome image, the pre- bidding
Label include disease species, disease severity grade or pathogen development rank.
In the present embodiment, the training image collection includes the image of the largely plant infected with disease, and the image is in
The symptom that existing plant shows by introduced disease.For example blade has withered and yellow spot, fruit appearance to have the diseases such as withered and yellow spot
Shape.The each image concentrated for training image, corresponding label, that is, default label can be set.The default label can
Including disease species and the severity of disease corresponding to the symptom reality that is presented in image.Understandably, training image collection
In the quantity of training subgraph can set as the case may be, be not especially limited here.
Step S620, using the training image collection, using deep learning identification model described in deep learning Algorithm for Training,
Deep learning identification model after being trained.
In the present embodiment, training image collection can be instructed using deep learning algorithm to deep learning identification model
Practice, with the deep learning identification model after being trained.Understandably, training image collection obtains all kinds of diseases pair by training
The characteristics of image answered, the characteristics of image can include one or more of rgb value, gray scale and texture.
Further, training image collection can be trained by models such as AlexNet, Vgg16, inception, with
Deep learning identification model after to training.Alternatively, the deep learning algorithm may be, but not limited to, convolutional neural networks
Algorithm, Recognition with Recurrent Neural Network algorithm, deep neural network algorithm etc., are not especially limited here.
Understandably, the deep learning identification model after training can be used for the characteristics of image for extracting images to be recognized, root
Matched according to the characteristics of image of extraction with the characteristics of image that above-mentioned training obtains, matching obtains images to be recognized and training subgraph
The similarity of picture, disease species, severity or pathogen development rank are just to treat corresponding to the maximum training subgraph of similarity
Identify disease species corresponding to image, severity or pathogen development rank.
Alternatively, the deep learning identification model after the images to be recognized input training by acquisition is identified
The step of classification, can include：By in spectrum-textural characteristics of the images to be recognized and the deep learning identification model
The spectrum textural characteristics for the training subgraph that training image is concentrated are matched, and obtain the images to be recognized and training
The similarity of image；Concentrated in the training image and choose the maximum training subgraph of similarity, the instruction that will be selected
Practice physical tags of the default label of subgraph as the disease of the images to be recognized, to obtain planting in the images to be recognized
Disease species, disease severity grade or the region of pathogen development rank and introduced disease of thing infection.
Wherein, the severity grade of the disease includes slight, partially gently, moderate, lays particular stress on.Wherein, severity grade can use
In the serious conditions for identifying introduced disease in images to be recognized.The severity grade can also pass through numeral or other literal tables
Show, such as can represent that severity becomes larger corresponding severity grade respectively with 1 to 5.Certainly, the severity grade may be used also
To be divided into the grade that other are different from mentioned kind, it is not especially limited here.
In the present embodiment, methods described can also include：Will be at least one corresponding to one species preset number plant
Images to be recognized is associated；For at least one images to be recognized corresponding to one species preset number plant, according to
Obtained disease species, disease severity grade or the region of pathogen development rank and introduced disease, calculates the species and presets
The disease index or pathogen spore maturation index of number plant, with obtain the disease index of the species preset number plant or
Pathogen spore maturation index.
Understandably, the image for having at least one region of disease by the plant infection obtained with species preset number is made
For corresponding images to be recognized, and at least one images to be recognized of acquisition is associated with the plants, with by this
Classification is identified in the associated images to be recognized of plants, obtains disease index of the plants in local introduced disease
Or pathogen spore maturation index.Wherein, the preset number can be configured according to actual conditions, not limited specifically here
It is fixed.
In the prior art, when judging the severity grade of disease, generally carried out by the personnel that observe and predict for having correlation experience
Identification judges.The disease that the difference personnel of observing and predicting correspond to identical severity may draw different severities, cause judged result not
One, it is unfavorable for carrying out early warning to plant disease.And the present invention is based on above-mentioned design, can unify to judge the severity grade of disease,
The accuracy and uniformity to disease severity grade classification can be improved, and then early warning timely and effectively can be carried out to plant disease
Prompting.In addition, being classified by server 300 to image recognition to be identified, human resources can be reduced, improve the effect to disease recognition
Rate and accuracy rate.
Step S650, corresponding early warning is sent according to the severity.
In the present embodiment, for the plant disease of different severities, different early warnings can be sent to make a distinction.
For example user terminal 100 is smart mobile phone, then server 300 can be sent in advance according to the severity that identification obtains to smart mobile phone
Alert prompt command, so that smart mobile phone sends early warning, its suggestion content may include to be identified for the images to be recognized
Disease species and disease severity.The mode of its early warning can include short message, phone.If have on the smart mobile phone with
The corresponding application program of plant disease identification early warning system 10, then corresponding early warning can be carried out by the application program and carried
Show.In addition, the severity can be indicated by corresponding numeral, and can divide to the numeral in different range, obtain
Corresponding severity grade described above.
Alternatively, methods described also includes：In advance by the disease species in the default label and corresponding disease control
Strategy is associated；Local current weather information is obtained, according to disease corresponding to the weather information and multiple images to be recognized
Severity calculate fashion trend of the disease in local preset time period；It is whole to the user according to the fashion trend
End 100 sends disease control strategy and/or early warning corresponding to the disease.Wherein, the multiple images to be recognized can be upper
State images to be recognized corresponding to the investigation sample (or same plants) of preset number.One investigation sample can be corresponding with least one
Individual images to be recognized.
Wherein, the weather information includes the future of local (such as one or more local field) relatively current time
Weather conditions (such as fine, rain etc.), outdoor in (being set for a period of time according to actual conditions, be not especially limited here)
Temperature data, humidity data etc., and the suitability under various circumstances of identified disease is combined, the disease will be obtained in future
The popular risk of a period of time, and corresponding early warning is sent, so that user prevents and treats the disease early.Understandably, it is described
Fashion trend can be regarded as the occurrence degree of the disease, can enter by using with the same or similar mode of the Severity gradation
Row represents, repeats no more here.
Further, methods described may also include：According to the disease species of identification, the strategy for preventing and treating the disease is obtained, with
The user terminal 100 is set to obtain the strategy.
In the present embodiment, server 300 can be by identifying obtained disease species, and acquisition prevents and treats the strategy of the disease,
To push the control strategy to for terminal, the reference policy to prevent and treat the disease as user, the experience sense of user is lifted.Separately
Outside, no user for preventing and treating the disease experience directly can be prevented and treated disease using the control strategy.
Fig. 5 is refer to, is the square frame signal for the plant disease identification prior-warning device 500 that present pre-ferred embodiments provide
Figure.Plant disease identification prior-warning device 500 can be used for above-mentioned plant disease identification method for early warning, to be carried out to plant disease
Classification and Identification, and send early warning.Plant disease identification prior-warning device 500 can include first acquisition unit 530, identification
Taxon 540 and early warning unit 550.
First acquisition unit 530, for obtaining the images to be recognized of colour, the images to be recognized includes corresponding light
Spectrum-textural characteristics.Specifically, first acquisition unit 530 can be used for performing the step S630 shown in Fig. 3, specific to perform
Content can refer to detailed description to step S630, repeat no more here.
Taxon 540 is identified, mould is identified for the deep learning after the images to be recognized input training by acquisition
Type is identified classification, and the deep learning identification model is according to spectrum-textural characteristics of the images to be recognized to obtain disease
The severity of evil species and disease.Specifically, identification taxon 540 can be used for performing the step S640 shown in Fig. 3, tool
The content of the execution of body can refer to the detailed description to step S640, repeat no more here.
Early warning unit 550, for sending corresponding early warning according to the severity.Specifically, early warning
Unit 550 can be used for performing the step S650 shown in Fig. 3, and the content specifically performed can refer to the detailed of step S650
Description, is repeated no more here.
Further, plant disease identification prior-warning device 500 can also include second acquisition unit 510 and model training
Unit 520.Second acquisition unit 510 and model training unit 520 can be held before identification taxon 540 performs step S640
Row corresponding contents.
Specifically, second acquisition unit 510, for obtaining the training image collection of plant disease, including multiple training subgraphs
Picture, each training subgraph include the plant image of introduced disease and the default label of corresponding disease species, wherein,
In leaf image of the plant image including plant, petal image, fruit image, limb/branch image, rhizome image extremely
Few one kind, the default label include disease species, disease severity grade or pathogen development rank.Specifically, second obtain
Unit 510 is taken to can be used for performing the step S610 shown in Fig. 4, the content specifically performed can refer to the detailed of step S610
Thin description, is repeated no more here.
Model training unit 520, for using the training image collection, using depth described in deep learning Algorithm for Training
Practise identification model, the deep learning identification model after being trained.Specifically, model training unit 520 can be used for performing Fig. 4
Shown in step S620, the content specifically performed can refer to the detailed description to step S620, repeats no more here.
Alternatively, the plant disease identification prior-warning device 500 goes back associative cell 560 and computing unit 570.
Specifically, associative cell 560, for by least one images to be recognized corresponding to one species preset number plant
It is associated.
Computing unit 570, for at least one images to be recognized corresponding to one species preset number plant,
According to obtained disease species, disease severity grade or the region of pathogen development rank and introduced disease, the species is calculated
The disease index or pathogen spore maturation index of preset number plant, are referred to obtaining the state of an illness of the species preset number plant
Number or pathogen spore maturation index.
In summary, the present invention provides a kind of plant disease identification method for early warning and device.This method and device are by obtaining
The images to be recognized of colour is taken, the images to be recognized includes corresponding spectrum-textural characteristics；By the figure to be identified of acquisition
Classification is identified in deep learning identification model after being trained as input, and the deep learning identification model is according to described to be identified
Spectrum-textural characteristics of image are to obtain the severity of disease species and disease；Corresponding early warning is sent according to the severity
Prompting.Scheme provided by the invention can reduce the manpower and material resources of identification plant disease, and improve the accurate of identification plant disease
Degree and recognition speed, in addition, the program also can obtain the severity of plant disease and send early warning according to the severity, have
Help related personnel and preventing and treating control is carried out to plant disease in time.
The preferred embodiments of the present invention are the foregoing is only, are not intended to limit the invention, for the skill of this area
For art personnel, the present invention can have various modifications and variations.Within the spirit and principles of the invention, that is made any repaiies
Change, equivalent substitution, improvement etc., should be included in the scope of the protection.
Claims (10)
1. a kind of plant disease identifies method for early warning, it is characterised in that identifies early warning system, the plant applied to plant disease
Disease recognition early warning system includes deep learning identification model, and methods described includes：
Colored images to be recognized is obtained, the images to be recognized includes corresponding spectrum-textural characteristics；
Classification, the depth is identified in deep learning identification model after the images to be recognized input training of acquisition
Identification model is practised according to spectrum-textural characteristics of the images to be recognized to obtain the severity of disease species and disease；
Corresponding early warning is sent according to the severity.
2. according to the method for claim 1, it is characterised in that after the images to be recognized input training by acquisition
Deep learning identification model the step of classification is identified before, methods described includes：
The training image collection of plant disease, including multiple training subgraphs are obtained, each training subgraph includes infection disease
Harmful plant image and the default label of corresponding disease species, wherein, leaf image of the plant image including plant,
At least one of petal image, fruit image, limb/branch image, rhizome image, the default label include disease kind
Class, disease severity grade or pathogen development rank；
Using the training image collection, using deep learning identification model described in deep learning Algorithm for Training, after being trained
Deep learning identification model.
3. according to the method for claim 2, it is characterised in that after the images to be recognized input training by acquisition
Deep learning identification model be identified classification, the deep learning identification model according to the spectrum of the images to be recognized-
The step of textural characteristics are to obtain the severity of disease species and disease, including：
The instruction that training image in spectrum-textural characteristics of the images to be recognized and the deep learning identification model is concentrated
The spectrum textural characteristics for practicing subgraph are matched, and obtain the images to be recognized and the similarity of the training subgraph；
Concentrated in the training image and choose the maximum training subgraph of similarity, by the training subgraph selected
Default physical tags of the label as the disease of the images to be recognized, to obtain the disease of plant infection in the images to be recognized
Evil species, disease severity grade or the region of pathogen development rank and introduced disease.
4. according to the method for claim 3, it is characterised in that methods described also includes：
At least one images to be recognized corresponding to one species preset number plant is associated；
For at least one images to be recognized corresponding to one species preset number plant, according to obtained disease species,
Disease severity grade or the region of pathogen development rank and introduced disease, the state of an illness for calculating the species preset number plant refer to
Number or pathogen spore maturation index, to obtain the disease index of the species preset number plant or pathogen spore maturity
Index.
5. according to the method for claim 2, it is characterised in that plant disease identification early warning system include server,
The user terminal being connected with the server communication and the image collecting device being connected with the server communication, the depth
Practise identification model to be arranged on the server, described image harvester is used for the coloured image for gathering the plant of introduced disease
To be sent as the images to be recognized, and by the images to be recognized to the deep learning identification model in the server.
6. according to the method for claim 5, it is characterised in that methods described also includes：
The disease species in the default label are associated with corresponding disease control strategy in advance；
Local current weather information is obtained, according to the serious of disease corresponding to the weather information and multiple images to be recognized
Degree calculates fashion trend of the disease in local preset time period；
According to the fashion trend, disease control strategy and/or early warning corresponding to the disease are sent to the user terminal.
7. a kind of plant disease identifies prior-warning device, it is characterised in that identifies early warning system, the plant applied to plant disease
Disease recognition early warning system includes deep learning identification model, and the plant disease identification prior-warning device includes：
First acquisition unit, for obtaining the images to be recognized of colour, it is special that the images to be recognized includes corresponding spectrum-texture
Sign；
Taxon is identified, is known for the deep learning identification model after the images to be recognized input training by acquisition
Do not classify, the deep learning identification model according to spectrum-textural characteristics of the images to be recognized with obtain disease species and
The severity of disease；
Early warning unit, for sending corresponding early warning according to the severity.
8. plant disease according to claim 7 identifies prior-warning device, it is characterised in that the plant disease identifies early warning
Device also includes：
Second acquisition unit, for obtaining the training image collection of plant disease, including multiple training subgraphs, each training
Subgraph includes the plant image of introduced disease and the default label of corresponding disease species, wherein, the plant image bag
At least one of the leaf image of plant, petal image, fruit image, limb/branch image, rhizome image are included, it is described pre-
Bidding label include disease species, disease severity grade or pathogen development rank；
Model training unit, for using the training image collection, identified using deep learning described in deep learning Algorithm for Training
Model, the deep learning identification model after being trained.
9. plant disease according to claim 8 identifies prior-warning device, it is characterised in that the identification taxon is also used
In：
The instruction that training image in spectrum-textural characteristics of the images to be recognized and the deep learning identification model is concentrated
The spectrum textural characteristics for practicing subgraph are matched, and obtain the images to be recognized and the similarity of the training subgraph；
Concentrated in the training image and choose the maximum training subgraph of similarity, by the training subgraph selected
Default physical tags of the label as the disease of the images to be recognized, to obtain the disease of plant infection in the images to be recognized
Evil species, disease severity grade or the region of pathogen development rank and introduced disease.
10. plant disease according to claim 7 identifies prior-warning device, it is characterised in that the plant disease identification is pre-
Alarm device also includes：
Associative cell, at least one images to be recognized corresponding to one species preset number plant to be associated；
Computing unit, for at least one images to be recognized corresponding to one species preset number plant, according to
Disease species, disease severity grade or the region of pathogen development rank and introduced disease arrived, calculate the species present count
The disease index or pathogen spore maturation index of mesh plant, to obtain the disease index or disease of the species preset number plant
Opportunistic pathogen spore maturation index.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710968700.3A CN107742290A (en) | 2017-10-18 | 2017-10-18 | Plant disease identifies method for early warning and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710968700.3A CN107742290A (en) | 2017-10-18 | 2017-10-18 | Plant disease identifies method for early warning and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107742290A true CN107742290A (en) | 2018-02-27 |
Family
ID=61237685
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710968700.3A Pending CN107742290A (en) | 2017-10-18 | 2017-10-18 | Plant disease identifies method for early warning and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107742290A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109145785A (en) * | 2018-08-03 | 2019-01-04 | 百度在线网络技术（北京）有限公司 | The determination method and apparatus of plant-nursing mode |
| CN109270909A (en) * | 2018-10-30 | 2019-01-25 | 四川洪诚电气科技有限公司 | Crops supervisory systems and method based on Internet of Things |
| CN109558883A (en) * | 2018-12-03 | 2019-04-02 | 宁夏智启连山科技有限公司 | Leaf characteristic extracting method and device |
| CN109754022A (en) * | 2019-01-24 | 2019-05-14 | 河池学院 | A kind of sugarcane plumule categorizing system based on convolutional neural networks |
| CN109801275A (en) * | 2019-01-11 | 2019-05-24 | 北京邮电大学 | Potato disease detection method and system based on image recognition |
| CN110251698A (en) * | 2018-12-27 | 2019-09-20 | 合刃科技（深圳）有限公司 | EO-1 hyperion Indoor Video method, device and equipment based on deep learning |
| CN111598181A (en) * | 2020-05-21 | 2020-08-28 | 中国热带农业科学院热带生物技术研究所 | Banana flower and leaf heart rot APP identification method and system |
| CN111967424A (en) * | 2020-08-27 | 2020-11-20 | 西南大学 | Buckwheat disease identification method based on convolutional neural network |
| CN111967441A (en) * | 2020-09-04 | 2020-11-20 | 郑州轻工业大学 | Crop disease analysis method based on deep learning |
| CN112465038A (en) * | 2020-11-30 | 2021-03-09 | 深圳市识农智能科技有限公司 | Method and system for identifying disease and insect pest types of fruit trees |
| CN113228047A (en) * | 2018-10-24 | 2021-08-06 | 克莱米特公司 | Plant disease detection using multi-stage, multi-scale deep learning |
| CN113297925A (en) * | 2021-04-30 | 2021-08-24 | 华中农业大学 | Intelligent early warning method and system for quality of full chain of fruits and vegetables |
| CN114829882A (en) * | 2019-11-05 | 2022-07-29 | 阿比尔技术公司 | Prediction of infection in plant products |
| CN115965875A (en) * | 2023-03-16 | 2023-04-14 | 德阳稷农农业科技有限公司 | Intelligent monitoring method and system for crop diseases and insect pests |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105825177A (en) * | 2016-03-09 | 2016-08-03 | 西安科技大学 | Remote-sensing crop disease identification method based on time phase and spectrum information and habitat condition |
| CN106022467A (en) * | 2016-05-17 | 2016-10-12 | 浪潮集团有限公司 | Crop disease detection system based on neural network |
| CN106202489A (en) * | 2016-07-20 | 2016-12-07 | 青岛云智环境数据管理有限公司 | A kind of agricultural pest intelligent diagnosis system based on big data |
| WO2017047814A1 (en) * | 2015-09-18 | 2017-03-23 | Ｐｓソリューションズ株式会社 | Image evaluation method |
| CN106971160A (en) * | 2017-03-23 | 2017-07-21 | 西京学院 | Winter jujube disease recognition method based on depth convolutional neural networks and disease geo-radar image |
-
2017
- 2017-10-18 CN CN201710968700.3A patent/CN107742290A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017047814A1 (en) * | 2015-09-18 | 2017-03-23 | Ｐｓソリューションズ株式会社 | Image evaluation method |
| CN105825177A (en) * | 2016-03-09 | 2016-08-03 | 西安科技大学 | Remote-sensing crop disease identification method based on time phase and spectrum information and habitat condition |
| CN106022467A (en) * | 2016-05-17 | 2016-10-12 | 浪潮集团有限公司 | Crop disease detection system based on neural network |
| CN106202489A (en) * | 2016-07-20 | 2016-12-07 | 青岛云智环境数据管理有限公司 | A kind of agricultural pest intelligent diagnosis system based on big data |
| CN106971160A (en) * | 2017-03-23 | 2017-07-21 | 西京学院 | Winter jujube disease recognition method based on depth convolutional neural networks and disease geo-radar image |
Non-Patent Citations (2)
| Title |
|---|
| 孙俊 等: ""基于改进卷积神经网络的多种植物叶片病害识别"", 《农业工程学报》 * |
| 王静 等: ""冬小麦条锈病严重度不同估算方法对比研究"", 《光谱学与光谱分析》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109145785A (en) * | 2018-08-03 | 2019-01-04 | 百度在线网络技术（北京）有限公司 | The determination method and apparatus of plant-nursing mode |
| CN113228047A (en) * | 2018-10-24 | 2021-08-06 | 克莱米特公司 | Plant disease detection using multi-stage, multi-scale deep learning |
| CN109270909A (en) * | 2018-10-30 | 2019-01-25 | 四川洪诚电气科技有限公司 | Crops supervisory systems and method based on Internet of Things |
| CN109558883A (en) * | 2018-12-03 | 2019-04-02 | 宁夏智启连山科技有限公司 | Leaf characteristic extracting method and device |
| CN109558883B (en) * | 2018-12-03 | 2023-04-18 | 宁夏智启连山科技有限公司 | Blade feature extraction method and device |
| CN110251698A (en) * | 2018-12-27 | 2019-09-20 | 合刃科技（深圳）有限公司 | EO-1 hyperion Indoor Video method, device and equipment based on deep learning |
| CN109801275A (en) * | 2019-01-11 | 2019-05-24 | 北京邮电大学 | Potato disease detection method and system based on image recognition |
| CN109754022A (en) * | 2019-01-24 | 2019-05-14 | 河池学院 | A kind of sugarcane plumule categorizing system based on convolutional neural networks |
| CN109754022B (en) * | 2019-01-24 | 2023-06-13 | 河池学院 | Sugarcane germ classification system based on convolutional neural network |
| CN114829882A (en) * | 2019-11-05 | 2022-07-29 | 阿比尔技术公司 | Prediction of infection in plant products |
| CN114829882B (en) * | 2019-11-05 | 2023-02-28 | 阿比尔技术公司 | Prediction of infection in plant products |
| CN111598181B (en) * | 2020-05-21 | 2023-04-07 | 中国热带农业科学院热带生物技术研究所 | Banana flower and leaf heart rot APP identification method and system |
| CN111598181A (en) * | 2020-05-21 | 2020-08-28 | 中国热带农业科学院热带生物技术研究所 | Banana flower and leaf heart rot APP identification method and system |
| CN111967424A (en) * | 2020-08-27 | 2020-11-20 | 西南大学 | Buckwheat disease identification method based on convolutional neural network |
| CN111967441A (en) * | 2020-09-04 | 2020-11-20 | 郑州轻工业大学 | Crop disease analysis method based on deep learning |
| CN112465038A (en) * | 2020-11-30 | 2021-03-09 | 深圳市识农智能科技有限公司 | Method and system for identifying disease and insect pest types of fruit trees |
| CN113297925A (en) * | 2021-04-30 | 2021-08-24 | 华中农业大学 | Intelligent early warning method and system for quality of full chain of fruits and vegetables |
| CN115965875A (en) * | 2023-03-16 | 2023-04-14 | 德阳稷农农业科技有限公司 | Intelligent monitoring method and system for crop diseases and insect pests |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107742290A (en) | Plant disease identifies method for early warning and device | |
| CN110148120B (en) | Intelligent disease identification method and system based on CNN and transfer learning | |
| JP6935377B2 (en) | Systems and methods for automatic inference of changes in spatiotemporal images | |
| CN104598908B (en) | A kind of crops leaf diseases recognition methods | |
| Yun et al. | PNN based crop disease recognition with leaf image features and meteorological data | |
| CN107463958A (en) | Insect identifies method for early warning and system | |
| CN111553240B (en) | Corn disease condition grading method and system and computer equipment | |
| CN111461165A (en) | Image recognition method, recognition model training method, related device and equipment | |
| Guo-feng et al. | A rapid, low-cost deep learning system to classify strawberry disease based on cloud service | |
| CN109344738A (en) | The recognition methods of crop diseases and pest crop smothering and device | |
| CN112949704B (en) | Tobacco leaf maturity state identification method and device based on image analysis | |
| CN106980852A (en) | Based on Corner Detection and the medicine identifying system matched and its recognition methods | |
| Mehta et al. | Improving Crop Health Management: Federated Learning CNN for Spinach Leaf Disease Detection | |
| CN114693616A (en) | Rice disease detection method, equipment and medium based on improved target detection model and convolutional neural network | |
| CN110874835B (en) | Crop leaf disease resistance identification method and system, electronic equipment and storage medium | |
| Sagarika et al. | Paddy plant disease classification and prediction using convolutional neural network | |
| Murugeswari et al. | Automated sugarcane disease detection using faster RCNN with an android application | |
| Tian et al. | Tomato leaf diseases recognition based on deep convolutional neural networks | |
| Desai et al. | A Novel technique for detecting crop diseases with efficient feature extraction | |
| CN206363347U (en) | Based on Corner Detection and the medicine identifying system that matches | |
| Mehta et al. | Apple Leaf Disease Recognition: A Robust Federated Learning CNN Methodology | |
| Chang et al. | Recognition of wheat rusts in a field environment based on improved DenseNet | |
| Pawar et al. | Detection of leaf diseases for multiple plants using convolutional neural network | |
| Sharma et al. | Detection and classification of plant diseases by Alexnet and GoogleNet deep learning architecture | |
| Mudgil et al. | Identification of Tomato Plant Diseases Using CNN-A Comparative Review |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180227 |