CN108427972A - Tobacco classification method based on on-line study and its system - Google Patents
Tobacco classification method based on on-line study and its system Download PDFInfo
- Publication number
- CN108427972A CN108427972A CN201810371929.3A CN201810371929A CN108427972A CN 108427972 A CN108427972 A CN 108427972A CN 201810371929 A CN201810371929 A CN 201810371929A CN 108427972 A CN108427972 A CN 108427972A
- Authority
- CN
- China
- Prior art keywords
- tobacco leaf
- tobacco
- leaf image
- classification
- data
- 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
- 235000002637 Nicotiana tabacum Nutrition 0.000 title claims abstract description 221
- 238000000034 method Methods 0.000 title claims abstract description 31
- 244000061176 Nicotiana tabacum Species 0.000 title 1
- 241000208125 Nicotiana Species 0.000 claims abstract description 220
- 238000013135 deep learning Methods 0.000 claims abstract description 43
- 230000008676 import Effects 0.000 claims abstract description 14
- 230000002708 enhancing effect Effects 0.000 claims description 27
- 238000012545 processing Methods 0.000 claims description 23
- 238000002372 labelling Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 2
- 238000012549 training Methods 0.000 description 13
- 238000005516 engineering process Methods 0.000 description 7
- 238000010801 machine learning Methods 0.000 description 7
- 230000003044 adaptive effect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000001965 increasing effect Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000012271 agricultural production Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 235000019504 cigarettes Nutrition 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000012797 qualification Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Classifications
-
- 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
- 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
Abstract
The present invention relates to tobacco classification method and its system based on on-line study, this method includes obtaining tobacco leaf image；Tobacco leaf image is handled, tobacco leaf image data are obtained；By treated, tobacco leaf image data import deep learning network；Tobacco classification is carried out using deep learning network.The present invention is by obtaining tobacco leaf image data and markup information, it is input to deep learning network Inception Network V3 as input terminal, in conjunction with its newly-increased depth network and sort module, obtain tobacco leaf feature, tobacco classification is carried out using tobacco leaf feature, realize efficient and high-accuracy tobacco classification, it is easy to operate, save cost.
Description
Technical field
The present invention relates to tobacco classification methods, more specifically refer to tobacco classification method based on on-line study and its are
System.
Background technology
Tobacco leaf crop occupies an important position in China's agricultural production.In the purchase of tobacco leaf crop, rating system is to peasant
It is of great significance with the interests of businessman.
China's tobacco leaf rating system and implementation rely primarily on the field monitoring in expert at present, have efficiency low, of high cost,
The obvious shortcomings such as subjectivity is strong, also, expert's classification is very high to manual request.The culture of expert needs stringent training mechanism,
The qualification of professional ability needs prolonged practical experience.In recent years, the rise with artificial intelligence in agricultural is applied, is based on
Automatic intelligent equipment and the technology of algorithm have found value with system in agricultural is applied, however, existing automatic tobacco leaf point
Grade system has many limitations, causes its effect in practical application that expected standard is not achieved, therefore there is still a need for a large amount of
Manpower checked, existing automatic classification system is mainly based upon the monokaryon categorizing system of traditional supervised learning, such
System has apparent limitation：Precision is low, traditional machine learning algorithm, such as the disaggregated model based on support vector machines, needs
Very important person is the design and extraction for carrying out feature, and based on the color artificially designed, the image classification of Texture eigenvalue is due to tobacco leaf figure
The complexity of picture, it is difficult to which the performance being optimal in practice is unable to fully efficiently be classified using all characteristics of image, mould
Type training speed is slow, and computational efficiency under conditions of big data of the machine learning training system based on monokaryon is low, leads to new mould
Type can not iteratively faster, adaptive ability is poor, and the existing machine learning training system based on batch can not rapidly adapt in a steady stream
The variation tendency of data flow constantly from practical operation place, it is single with becoming larger for data volume and increasing for new varieties
Core processor needs the new disaggregated model of training of starting from scratch, inefficiency, it is also possible to due to that can not handle all numbers simultaneously
Fail according to model training is caused.
Therefore, it is necessary to be designed with kind of a new tobacco classification method, efficient and high-accuracy tobacco classification, operation are realized
Simply, cost is saved.
Invention content
It is an object of the invention to overcome the deficiencies of existing technologies, provide tobacco classification method based on on-line study and its
System.
To achieve the above object, the present invention uses following technical scheme：Tobacco classification method based on on-line study, it is described
Method includes：
Obtain tobacco leaf image；
Tobacco leaf image is handled, tobacco leaf image data are obtained；
By treated, tobacco leaf image data import deep learning network；
Tobacco classification is carried out using deep learning network.
Its further technical solution is：The step of tobacco leaf image is handled, including step in detail below：
Tobacco leaf image is labeled, markup information is obtained；
Enhancing processing is carried out to tobacco leaf image.
Its further technical solution is：The step of enhancing processing is carried out to tobacco leaf image, including step in detail below：
Adjust the shooting angle of tobacco leaf image；
Adjust the scaling of tobacco leaf image；
The position of tobacco leaf is adjusted, tobacco leaf image data are formed.
Its further technical solution is：The step of tobacco classification being carried out using deep learning network, including walk in detail below
Suddenly：
The fine tuning for carrying out parameter to the top layer of deep learning network using tobacco leaf image data, the tobacco leaf for obtaining tobacco leaf data are special
Sign；
Tobacco classification is carried out according to tobacco leaf feature.
Its further technical solution is：After the step of carrying out tobacco classification according to tobacco leaf feature, further include：
Using tobacco leaf image data as input data, the parameter update of depth network is carried out, and carries out data enhancing, is generated various
Change data.
The present invention also provides the tobacco classification system based on on-line study, including image acquisition unit, processing unit, lead
Enter unit and taxon；
Described image acquiring unit, for obtaining tobacco leaf image；
The processing unit obtains tobacco leaf image data for handling tobacco leaf image；
The import unit, for tobacco leaf image data to import deep learning network by treated；
The taxon, for carrying out tobacco classification using deep learning network.
Its further technical solution is：The processing unit includes labeling module and enhancing module；
The labeling module obtains markup information for being labeled to tobacco leaf image；
The enhancing module, for carrying out enhancing processing to tobacco leaf image.
Its further technical solution is：The enhancing module further include angle adjustment submodule, ratio adjustment submodule with
And position adjustment submodule；
The angle adjusts submodule, the shooting angle for adjusting tobacco leaf image；
The ratio adjusts submodule, the scaling for adjusting tobacco leaf image；
The position adjustment submodule, the position for adjusting tobacco leaf form tobacco leaf image data.
Its further technical solution is：The taxon includes fine tuning module and tobacco classification module；
The fine tuning module, the fine tuning for being carried out parameter to the top layer of deep learning network using tobacco leaf image data are obtained
The tobacco leaf feature of tobacco leaf data；
The tobacco classification module, for carrying out tobacco classification according to tobacco leaf feature.
Its further technical solution is：The taxon further includes update module；
The update module for using tobacco leaf image data as input data, carrying out the parameter update of depth network, and carries out
Data enhance, and generate enriched data.
Compared with the prior art, the invention has the advantages that：The tobacco classification method based on on-line study of the present invention,
By obtaining tobacco leaf image data and markup information, deep learning network Inception is input to as input terminal
Network V3 obtain tobacco leaf feature in conjunction with its newly-increased depth network and sort module, and tobacco leaf is carried out using tobacco leaf feature
Efficient and high-accuracy tobacco classification is realized in classification, easy to operate, saves cost.
The invention will be further described in the following with reference to the drawings and specific embodiments.
Description of the drawings
Fig. 1 is the flow chart one for the tobacco classification method based on on-line study that the specific embodiment of the invention provides；
Fig. 2 is the flowchart 2 for the tobacco classification method based on on-line study that the specific embodiment of the invention provides；
Fig. 3 is the flow chart handled tobacco leaf image that the specific embodiment of the invention provides；
Fig. 4 is the flow chart that enhancing processing is carried out to tobacco leaf image that the specific embodiment of the invention provides；
Fig. 5 is the flow chart that tobacco classification is carried out using deep learning network that the specific embodiment of the invention provides；
Fig. 6 is the newer flow chart of parameter that the specific embodiment of the invention provides；
Fig. 7 is the structure diagram for the deep learning network that the specific embodiment of the invention provides；
Fig. 8 is the structure diagram for the tobacco classification system based on on-line study that the specific embodiment of the invention provides；
Fig. 9 is the structure diagram for the processing unit that the specific embodiment of the invention provides；
Figure 10 is the structure diagram for the enhancing module that the specific embodiment of the invention provides；
Figure 11 is the structure diagram for the taxon that the specific embodiment of the invention provides.
Specific implementation mode
In order to more fully understand the present invention technology contents, with reference to specific embodiment to technical scheme of the present invention into
One step introduction and explanation, but not limited to this.
The specific embodiment as shown in Fig. 1~11, the tobacco classification method provided in this embodiment based on on-line study, can
To apply to the classification of tobacco leaf and the extraction of feature, realize efficiently and the tobacco classification of high-accuracy, it is easy to operate, save at
This.
As shown in Figure 1, present embodiments providing the tobacco classification method based on on-line study, this method includes：
S1, tobacco leaf image is obtained；
S2, tobacco leaf image is handled, obtains tobacco leaf image data；
S3, by treated, tobacco leaf image data import deep learning network；
S4, tobacco classification is carried out using deep learning network.
For above-mentioned S1 steps, tobacco leaf image is obtained using camera shooting tobacco leaf especially by professional person.
Further, in certain embodiments, above-mentioned S2 steps, the step of processing tobacco leaf image, including
Step in detail below：
S21, tobacco leaf image is labeled, obtains markup information；
S22, enhancing processing is carried out to tobacco leaf image.
For above-mentioned S21 step, the new new tobacco leaf image that do not mark is labeled, markup information is obtained,
Using markup information and tobacco leaf image as the input terminal of deep learning network.
For above-mentioned S22 steps, the step of enhancing is handled, including step in detail below are carried out to tobacco leaf image：
S221, the shooting angle for adjusting tobacco leaf image；
S222, the scaling for adjusting tobacco leaf image；
S223, the position for adjusting tobacco leaf form tobacco leaf image data.
Above-mentioned S221 steps to S223 steps can effectively be increased based on the data acquisition scheme of data enhancing
Add data volume, prevent the overfitting phenomenon to training data occurred in machine-learning process, is actually being answered to increase
With the accuracy rate in stage.
For above-mentioned S3 steps, specifically tobacco leaf image data and markup information are directed into based on thousands of class figures
As the deep learning network Inception Network V3 of classification and depth network module newly-increased on it and classification
In module, the acquisition of feature and the classification of tobacco leaf are carried out.
Further, for above-mentioned S4 steps, the step of carrying out tobacco classification using deep learning network, including with
Lower specific steps：
S41, the fine tuning for carrying out parameter to the top layer of deep learning network using tobacco leaf image data, obtain the tobacco leaf of tobacco leaf data
Feature；
S42, tobacco classification is carried out according to tobacco leaf feature.
For above-mentioned S41 steps, tobacco leaf feature include in residual degree of wound, the shape of tobacco leaf, area, perimeter, circularity at least
One, it can not only efficiently use publicly-owned mass data（Imagenet databases）Learn with generality rudimentary and
Mid-level features can also obtain the advanced features for being absorbed in tobacco leaf data with deep learning.In addition, above-mentioned deep learning network
By data flow carry out small parameter perturbations formula be：
；
；W and b is the parameter of deep learning network model, and wherein b is per layer network
Intercept, η are learning rate, and m is the amount of training data（The namely amount of tobacco leaf image data）, C is object function, passes through gradient
Descent algorithm to existing parameter w and b only depend on the fine tuning of current new input data.
In addition, for above-mentioned S42 steps, one kind is attributed to according to what the numerical value of tobacco leaf feature met setting range.
Further, above-mentioned S42 steps, according to tobacco leaf feature carry out tobacco classification the step of after, further include：
S43, using tobacco leaf image data as input data, carry out the parameter update of depth network, and carry out data enhancing, generate
Enriched data.
It specifically,, can be automatically by data after obtaining a certain number of new mark tobacco leaf datas for above-mentioned S43 steps
The parameter update of depth network is carried out as input, and carries out data enhancing and generates more diversified data, to prevent model
The overfitting of training system on the training data efficiently carries out the model modification based on high amount of traffic to realize.It is traditional deep
The tobacco leaf image that degree learning network expert team newly marks can be automatically uploaded in the stocking system based on high in the clouds, for other classification
System is called.
This method with deep learning network structure as shown in fig. 7, full automatic data chain system gets through life
Producing line expert marks and high in the clouds machine learning system, has classification effectiveness accuracy rate high, and adaptive new data ability is strong, operation
Simply, a large amount of human resources are saved；The full-automatic adaptive learning network in high in the clouds, Neng Gou are serviced based on on-line study and network
The tobacco leaf data stream with new varieties or new appearance is rapidly adapted in time in minutes.
The above-mentioned tobacco classification method based on on-line study is made by obtaining tobacco leaf image data and markup information
Deep learning network Inception Network V3 are input to for input terminal, in conjunction with its newly-increased depth network and classification mould
Block obtains tobacco leaf feature, and tobacco classification is carried out using tobacco leaf feature, realizes efficient and high-accuracy tobacco classification, operation letter
It is single, save cost.
As shown in figure 8, the present invention also provides the tobacco classification systems based on on-line study comprising image acquisition unit
1, processing unit 2, import unit 3 and taxon 4.
Image acquisition unit 1, for obtaining tobacco leaf image.It is obtained using camera shooting tobacco leaf especially by professional person
Tobacco leaf image.
Processing unit 2 obtains tobacco leaf image data for handling tobacco leaf image.
Import unit 3, for tobacco leaf image data to import deep learning network by treated.
Taxon 4, for carrying out tobacco classification using deep learning network.
Further, in certain embodiments, above-mentioned processing unit 2 includes labeling module 21 and enhancing module
22。
Labeling module 21 obtains markup information for being labeled to tobacco leaf image.To new new tobacco leaf figure is not marked
As being labeled, markup information is obtained, using markup information and tobacco leaf image as the input terminal of deep learning network.
Enhance module 22, for carrying out enhancing processing to tobacco leaf image.
For above-mentioned enhancing module 22, enhancing module 22 further includes angle adjustment submodule 221, ratio adjustment
Module 222 and position adjustment submodule 223.
Angle adjusts submodule 221, the shooting angle for adjusting tobacco leaf image.
Ratio adjusts submodule 222, the scaling for adjusting tobacco leaf image.
Position adjustment submodule 223, the position for adjusting tobacco leaf form tobacco leaf image data.
Based on the data acquisition scheme of data enhancing, it can effectively increase data volume, prevent in machine-learning process
The overfitting phenomenon to training data occurred, to increase the accuracy rate in practical stage.
For above-mentioned import unit 3, specifically tobacco leaf image data and markup information are directed into and are based on
The deep learning network Inception Network V3 of thousands of class image classifications and depth network module newly-increased on it
And in sort module, the acquisition of feature and the classification of tobacco leaf are carried out.
In addition, in certain embodiments, above-mentioned taxon 4 includes fine tuning module 41 and tobacco classification module 42.
Module 41 is finely tuned, the fine tuning for being carried out parameter to the top layer of deep learning network using tobacco leaf image data is obtained
Take the tobacco leaf feature of tobacco leaf data.Tobacco leaf feature include it is at least one in residual degree of wound, the shape of tobacco leaf, area, perimeter, circularity,
Publicly-owned mass data can not only be efficiently used（Imagenet databases）What is learnt is rudimentary and intermediate with generality
Feature can also obtain the advanced features for being absorbed in tobacco leaf data with deep learning.In addition, above-mentioned deep learning network passes through
Data flow carry out small parameter perturbations formula be：
W and b is the parameter of deep learning network model, and wherein b is the intercept per layer network, and η is learning rate, and m is trained number
According to amount（The namely amount of tobacco leaf image data）, C is object function, by gradient descent algorithm to existing parameter w and b into
Row only depends on the fine tuning of current new input data.
Tobacco classification module 42, for carrying out tobacco classification according to tobacco leaf feature.It is set according to the numerical value of tobacco leaf feature satisfaction
That determines range is attributed to one kind.
Further, above-mentioned taxon 4 further includes update module 43；
The update module 43, for using tobacco leaf image data as input data, carrying out the parameter update of depth network, going forward side by side
Row data enhance, and generate enriched data.
After obtaining a certain number of new mark tobacco leaf datas, the parameter of depth network can be carried out using data as input automatically
Update, and carry out data enhancing and generate more diversified data, to prevent model training systems on the training data excessive
Fitting efficiently carries out the model modification based on high amount of traffic to realize.Conventional depth learning network expert team newly marks
Tobacco leaf image can be automatically uploaded in the stocking system based on high in the clouds, be called for other categorizing systems.
This method with deep learning network structure as shown in fig. 7, full automatic data chain system gets through life
Producing line expert marks and high in the clouds machine learning system, has classification effectiveness accuracy rate high, and adaptive new data ability is strong, operation
Simply, a large amount of human resources are saved；The full-automatic adaptive learning network in high in the clouds, Neng Gou are serviced based on on-line study and network
The tobacco leaf data stream with new varieties or new appearance is rapidly adapted in time in minutes.
The embodiment of the present invention additionally provides the tobacco classification system based on on-line study comprising：One or more processing
Device, memory, and, one or more programs, the one or more program is stored in memory, and being configured as can be by institute
It states processor and reads execution, one or more of programs include the instruction that can be used for executing following steps：
Obtain tobacco leaf image；
Tobacco leaf image is handled, tobacco leaf image data are obtained；
By treated, tobacco leaf image data import deep learning network；
Tobacco classification is carried out using deep learning network.
The processor further includes the cigarette based on on-line study of any application described in above method embodiment when executing
Some or all of leaf sorting technique step.
The above-mentioned tobacco classification system based on on-line study is made by obtaining tobacco leaf image data and markup information
Deep learning network Inception Network V3 are input to for input terminal, in conjunction with its newly-increased depth network and classification mould
Block obtains tobacco leaf feature, and tobacco classification is carried out using tobacco leaf feature, realizes efficient and high-accuracy tobacco classification, operation letter
It is single, save cost.
It should be noted that for each method embodiment above-mentioned, for simple description, therefore it is all expressed as a series of
Combination of actions, but those skilled in the art should understand that, the present invention is not limited by the described action sequence because
According to the present invention, certain steps can be performed in other orders or simultaneously.Secondly, those skilled in the art should also know
It knows, embodiment described in this description belongs to preferred embodiment, and involved action and module are not necessarily of the invention
It is necessary.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, there is no the portion being described in detail in some embodiment
Point, it may refer to the associated description of other embodiment.
In several embodiments provided herein, it should be understood that disclosed device, it can be by another way
It realizes.For example, the apparatus embodiments described above are merely exemplary, for example, the unit division, it is only a kind of
Division of logic function, formula that in actual implementation, there may be another division manner, such as multiple units or component can combine or can
To be integrated into another system, or some features can be ignored or not executed.
The unit illustrated as separating component may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, you can be located at a place, or may be distributed over multiple
In network element.Some or all of unit therein can be selected according to the actual needs to realize the mesh of this embodiment scheme
's.
In addition, each functional unit in various embodiments of the present invention can be integrated in a processing unit, also may be used
It, can also be during two or more units be integrated in one unit to be that each unit physically exists alone.It is above-mentioned integrated
The form that hardware had both may be used in unit is realized, can also be realized in the form of SFU software functional unit.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a processor, i.e., in computer read/write memory medium.Based on this understanding, skill of the invention
Substantially all or part of the part that contributes to existing technology or the technical solution can be with soft in other words for art scheme
The form of part product embodies, which is stored in a storage medium, including some instructions are making
Obtain a computer equipment（Can be personal computer, server or network equipment etc.）It executes described in each embodiment of the present invention
The all or part of step of method.And storage medium above-mentioned includes：USB flash disk, read-only memory（ROM, Read-Only
Memory）, random access memory（RAM, Random Access Memory）, mobile hard disk, magnetic disc or CD etc. it is various
The medium of program code can be stored.
It is above-mentioned only with embodiment come the technology contents that further illustrate the present invention, in order to which reader is easier to understand, but not
It represents embodiments of the present invention and is only limitted to this, any technology done according to the present invention extends or recreation, by the present invention's
Protection.Protection scope of the present invention is subject to claims.
Claims (10)
1. the tobacco classification method based on on-line study, which is characterized in that the method includes：
Obtain tobacco leaf image；
Tobacco leaf image is handled, tobacco leaf image data are obtained；
By treated, tobacco leaf image data import deep learning network；
Tobacco classification is carried out using deep learning network.
2. the tobacco classification method according to claim 1 based on on-line study, which is characterized in that carried out to tobacco leaf image
The step of processing, including step in detail below：
Tobacco leaf image is labeled, markup information is obtained；
Enhancing processing is carried out to tobacco leaf image.
3. the tobacco classification method according to claim 2 based on on-line study, which is characterized in that carried out to tobacco leaf image
The step of enhancing processing, including step in detail below：
Adjust the shooting angle of tobacco leaf image；
Adjust the scaling of tobacco leaf image；
The position of tobacco leaf is adjusted, tobacco leaf image data are formed.
4. the tobacco classification method according to any one of claims 1 to 3 based on on-line study, which is characterized in that utilize
Deep learning network carries out the step of tobacco classification, including step in detail below：
The fine tuning for carrying out parameter to the top layer of deep learning network using tobacco leaf image data, the tobacco leaf for obtaining tobacco leaf data are special
Sign；
Tobacco classification is carried out according to tobacco leaf feature.
5. the tobacco classification method according to claim 4 based on on-line study, which is characterized in that according to tobacco leaf feature into
After the step of row tobacco classification, further include：
Using tobacco leaf image data as input data, the parameter update of depth network is carried out, and carries out data enhancing, is generated various
Change data.
6. the tobacco classification system based on on-line study, which is characterized in that including image acquisition unit, processing unit, import list
Member and taxon；
Described image acquiring unit, for obtaining tobacco leaf image；
The processing unit obtains tobacco leaf image data for handling tobacco leaf image；
The import unit, for tobacco leaf image data to import deep learning network by treated；
The taxon, for carrying out tobacco classification using deep learning network.
7. the tobacco classification system according to claim 6 based on on-line study, which is characterized in that the processing unit packet
Include labeling module and enhancing module；
The labeling module obtains markup information for being labeled to tobacco leaf image；
The enhancing module, for carrying out enhancing processing to tobacco leaf image.
8. the tobacco classification system according to claim 7 based on on-line study, which is characterized in that the enhancing module is also
Including angle adjustment submodule, ratio adjustment submodule and position adjustment submodule；
The angle adjusts submodule, the shooting angle for adjusting tobacco leaf image；
The ratio adjusts submodule, the scaling for adjusting tobacco leaf image；
The position adjustment submodule, the position for adjusting tobacco leaf form tobacco leaf image data.
9. the tobacco classification system according to claim 8 based on on-line study, which is characterized in that the taxon packet
Include fine tuning module and tobacco classification module；
The fine tuning module, the fine tuning for being carried out parameter to the top layer of deep learning network using tobacco leaf image data are obtained
The tobacco leaf feature of tobacco leaf data；
The tobacco classification module, for carrying out tobacco classification according to tobacco leaf feature.
10. the tobacco classification system according to claim 9 based on on-line study, which is characterized in that the taxon
It further include update module；
The update module for using tobacco leaf image data as input data, carrying out the parameter update of depth network, and carries out
Data enhance, and generate enriched data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810371929.3A CN108427972A (en) | 2018-04-24 | 2018-04-24 | Tobacco classification method based on on-line study and its system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810371929.3A CN108427972A (en) | 2018-04-24 | 2018-04-24 | Tobacco classification method based on on-line study and its system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108427972A true CN108427972A (en) | 2018-08-21 |
Family
ID=63161541
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810371929.3A Pending CN108427972A (en) | 2018-04-24 | 2018-04-24 | Tobacco classification method based on on-line study and its system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108427972A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113469233A (en) * | 2021-06-23 | 2021-10-01 | 临沂大学 | Tobacco leaf automatic grading method and system based on deep learning |
Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102590211A (en) * | 2011-01-11 | 2012-07-18 | 郑州大学 | Method for utilizing spectral and image characteristics to grade tobacco leaves |
| CN103323455A (en) * | 2013-04-24 | 2013-09-25 | 南京文采科技有限责任公司 | Tobacco leaf grading method based on reflection, perspective and microscopic images |
| CN103743486A (en) * | 2014-01-02 | 2014-04-23 | 上海大学 | Automatic grading system and method based on mass tobacco leaf data |
| CN104573736A (en) * | 2015-01-20 | 2015-04-29 | 湖南中烟工业有限责任公司 | Tobacco stem and tobacco leaf precise recognition method and system based on dual-energy induction |
| CN105069472A (en) * | 2015-08-03 | 2015-11-18 | 电子科技大学 | Vehicle detection method based on convolutional neural network self-adaption |
| CN106250896A (en) * | 2016-03-03 | 2016-12-21 | 上海创和亿电子科技发展有限公司 | The recognition methods of the positive and negative of online Nicotiana tabacum L. based on image collecting device |
| CN106295584A (en) * | 2016-08-16 | 2017-01-04 | 深圳云天励飞技术有限公司 | Depth migration study is in the recognition methods of crowd's attribute |
| CN106326899A (en) * | 2016-08-18 | 2017-01-11 | 郑州大学 | Tobacco leaf grading method based on hyperspectral image and deep learning algorithm |
| CN106682694A (en) * | 2016-12-27 | 2017-05-17 | 复旦大学 | Sensitive image identification method based on depth learning |
| CN106991439A (en) * | 2017-03-28 | 2017-07-28 | 南京天数信息科技有限公司 | Image-recognizing method based on deep learning and transfer learning |
| CN107016413A (en) * | 2017-03-31 | 2017-08-04 | 征图新视（江苏）科技有限公司 | A kind of online stage division of tobacco leaf based on deep learning algorithm |
| CN107203740A (en) * | 2017-04-24 | 2017-09-26 | 华侨大学 | A kind of face age estimation method based on deep learning |
| CN107305636A (en) * | 2016-04-22 | 2017-10-31 | 株式会社日立制作所 | Target identification method, Target Identification Unit, terminal device and target identification system |
| CN107633255A (en) * | 2017-08-11 | 2018-01-26 | 天津大学 | A kind of rock lithology automatic recognition classification method under deep learning pattern |
| CN107679525A (en) * | 2017-11-01 | 2018-02-09 | 腾讯科技（深圳）有限公司 | Image classification method, device and computer-readable recording medium |
| CN107909566A (en) * | 2017-10-28 | 2018-04-13 | 杭州电子科技大学 | A kind of image-recognizing method of the cutaneum carcinoma melanoma based on deep learning |
| CN107945182A (en) * | 2018-01-02 | 2018-04-20 | 东北农业大学 | Maize leaf disease recognition method based on convolutional neural networks model GoogleNet |
-
2018
- 2018-04-24 CN CN201810371929.3A patent/CN108427972A/en active Pending
Patent Citations (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102590211A (en) * | 2011-01-11 | 2012-07-18 | 郑州大学 | Method for utilizing spectral and image characteristics to grade tobacco leaves |
| CN103323455A (en) * | 2013-04-24 | 2013-09-25 | 南京文采科技有限责任公司 | Tobacco leaf grading method based on reflection, perspective and microscopic images |
| CN103743486A (en) * | 2014-01-02 | 2014-04-23 | 上海大学 | Automatic grading system and method based on mass tobacco leaf data |
| CN104573736A (en) * | 2015-01-20 | 2015-04-29 | 湖南中烟工业有限责任公司 | Tobacco stem and tobacco leaf precise recognition method and system based on dual-energy induction |
| CN105069472A (en) * | 2015-08-03 | 2015-11-18 | 电子科技大学 | Vehicle detection method based on convolutional neural network self-adaption |
| CN106250896A (en) * | 2016-03-03 | 2016-12-21 | 上海创和亿电子科技发展有限公司 | The recognition methods of the positive and negative of online Nicotiana tabacum L. based on image collecting device |
| CN107305636A (en) * | 2016-04-22 | 2017-10-31 | 株式会社日立制作所 | Target identification method, Target Identification Unit, terminal device and target identification system |
| CN106295584A (en) * | 2016-08-16 | 2017-01-04 | 深圳云天励飞技术有限公司 | Depth migration study is in the recognition methods of crowd's attribute |
| CN106326899A (en) * | 2016-08-18 | 2017-01-11 | 郑州大学 | Tobacco leaf grading method based on hyperspectral image and deep learning algorithm |
| CN106682694A (en) * | 2016-12-27 | 2017-05-17 | 复旦大学 | Sensitive image identification method based on depth learning |
| CN106991439A (en) * | 2017-03-28 | 2017-07-28 | 南京天数信息科技有限公司 | Image-recognizing method based on deep learning and transfer learning |
| CN107016413A (en) * | 2017-03-31 | 2017-08-04 | 征图新视（江苏）科技有限公司 | A kind of online stage division of tobacco leaf based on deep learning algorithm |
| CN107203740A (en) * | 2017-04-24 | 2017-09-26 | 华侨大学 | A kind of face age estimation method based on deep learning |
| CN107633255A (en) * | 2017-08-11 | 2018-01-26 | 天津大学 | A kind of rock lithology automatic recognition classification method under deep learning pattern |
| CN107909566A (en) * | 2017-10-28 | 2018-04-13 | 杭州电子科技大学 | A kind of image-recognizing method of the cutaneum carcinoma melanoma based on deep learning |
| CN107679525A (en) * | 2017-11-01 | 2018-02-09 | 腾讯科技（深圳）有限公司 | Image classification method, device and computer-readable recording medium |
| CN107945182A (en) * | 2018-01-02 | 2018-04-20 | 东北农业大学 | Maize leaf disease recognition method based on convolutional neural networks model GoogleNet |
Non-Patent Citations (5)
| Title |
|---|
| MOSTAFA MEHDIPOUR GHAZI 等: "Plant identification using deep neural networks via optimization of transfer learning parameters", 《NEUROCOMPUTING》, pages 228 - 235 * |
| SIVA KRISHNA DASARI 等: "Flue-Cured Tobacco Leaves Classification: A Generalized Approach Using Deep Convolutional Neural Networks", pages 13 - 21 * |
| ZHANG JIANQIANG 等: "Sparse Representation Classification of Tobacco Leaves Using near-Infrared Spectroscopy and a Deep Learning Algorithm", pages 1 - 20 * |
| 赵媛 等: "一种基于深度学习的颈动脉斑块超声图像识别方法", 《中国医疗器械信息》, pages 9 - 11 * |
| 韩力群 等: "烤烟烟叶自动分级的智能技术", vol. 18, no. 6, pages 173 - 175 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113469233A (en) * | 2021-06-23 | 2021-10-01 | 临沂大学 | Tobacco leaf automatic grading method and system based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107145503A (en) | Remote supervision non-categorical relation extracting method and system based on word2vec | |
| CN107203891A (en) | A kind of automatic many threshold values characteristic filter method and devices | |
| CN110046952A (en) | A kind of training method and device, a kind of recommended method and device of recommended models | |
| CN107766787A (en) | Face character recognition methods, device, terminal and storage medium | |
| Lee et al. | High efficiency disease detection for potato leaf with convolutional neural network | |
| CN106980873A (en) | Fancy carp screening technique and device based on deep learning | |
| CN110135231A (en) | Animal face recognition methods, device, computer equipment and storage medium | |
| CN110083728B (en) | Method, device and system for optimizing automatic picture data cleaning quality | |
| CN110750567A (en) | Wearing and matching data generation method and system based on intelligent wardrobe | |
| CN112001242B (en) | Intelligent gardening management method and device | |
| CN110569883B (en) | Air quality index prediction method based on Kohonen network clustering and Relieff feature selection | |
| CN110059677A (en) | Digital table recognition methods and equipment based on deep learning | |
| CN108427972A (en) | Tobacco classification method based on on-line study and its system | |
| Ma et al. | Research on fish image classification based on transfer learning and convolutional neural network model | |
| Qi et al. | Related study based on otsu watershed algorithm and new squeeze-and-excitation networks for segmentation and level classification of tea buds | |
| Yaiprasert | Artificial intelligence for para rubber identification combining five machine learning methods | |
| CN111191527B (en) | Attribute identification method, attribute identification device, electronic equipment and readable storage medium | |
| CN116612435B (en) | Corn high-yield cultivation method | |
| CN113313692A (en) | Automatic banana young plant identification and counting method based on aerial visible light image | |
| CN103793714B (en) | Many Classification and Identification device generating means and its method, data identification means and its method | |
| CN113221913A (en) | Agriculture and forestry disease and pest fine-grained identification method and device based on Gaussian probability decision-level fusion | |
| CN112446417A (en) | Spindle-shaped fruit image segmentation method and system based on multilayer superpixel segmentation | |
| CN110163224B (en) | Auxiliary data labeling method capable of online learning | |
| CN109635764A (en) | A kind of Human bodys' response method and system based on multiple features linear temporal coding | |
| CN113128514A (en) | Cotton pest positioning and classifying identification method and system |
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 |