CN110275974A - Data adding method, device, terminal device and the storage medium of sample data set - Google Patents
Data adding method, device, terminal device and the storage medium of sample data set Download PDFInfo
- Publication number
- CN110275974A CN110275974A CN201910583761.7A CN201910583761A CN110275974A CN 110275974 A CN110275974 A CN 110275974A CN 201910583761 A CN201910583761 A CN 201910583761A CN 110275974 A CN110275974 A CN 110275974A
- Authority
- CN
- China
- Prior art keywords
- picture
- data set
- added
- weighted value
- sample 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/55—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/955—Retrieval from the web using information identifiers, e.g. uniform resource locators [URL]
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Library & Information Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses data adding method, device, terminal device and the storage mediums of a kind of sample data set, comprising: obtains the picture classification of sample data set to be added;Grab picture to be selected corresponding with the picture classification;The fisrt feature weighted value of each picture to be selected is determined by default neural network model, and obtains the second feature weighted value that the sample data to be added concentrates storage;Picture to be added is chosen from the picture to be selected according to the fisrt feature weighted value and the second feature weighted value;The picture to be added is added to the sample data set to be added, the present invention allows neural network to extract the more features of same class picture from more pictures, to reach the addition to data set and effect of optimization by constantly updating to data.
Description
Technical field
The invention belongs to deep learning fields, are related to data adding method, device, the terminal device of a kind of sample data set
And storage medium.
Background technique
Deep learning is because it has good effect to the extraction of feature, so the identification and classification in image have extensively
Application.Because the effect of the classification of image has very high correlation to feature extraction degree, in practical applications, for
Image classification problem is generally adopted by and extracts feature using neural network.The data set used in neural network is to training
Result also have very big influence, this is because neural network be the weight of different characteristic is obtained by data set, so
To the collection of data set also important in inhibiting.For image classification, for the accuracy rate and knowledge of neural network recognization classification
Other speed has become for two important indicators.In practical application, operand during identification, the size of data set,
The complexity of model suffers from very big influence to last result.
Presently most common method be using convolutional neural networks extract feature, but due to data set be it is static,
It is changeless, cause after neural metwork training is complete, is difficult again to update data set, thus to the weight of feature after training
Also it can not advanced optimize.And traditional data set is fixed and invariable, and data set also accounts for when being put into training in neural network
According to very big memory, and then affect finally trained speed and result.
Above content is only used to facilitate the understanding of the technical scheme, and is not represented and is recognized that above content is existing skill
Art.
Summary of the invention
The main purpose of the present invention is to provide data adding method, device, the terminal devices of a kind of sample data set
And storage medium, it is intended to solve the high cost of the prior art and macrocyclic technical problem.
To achieve the above object, the present invention provides a kind of data adding method of sample data set, the method includes
Following steps:
Obtain the picture classification of sample data set to be added;
Grab picture to be selected corresponding with the picture classification;
The fisrt feature weighted value of each picture to be selected is determined by default neural network model, and obtains the sample to be added
The centrally stored second feature weighted value of notebook data;
It is chosen from the picture to be selected according to the fisrt feature weighted value and the second feature weighted value to be added
Picture;
The picture to be added is added to the sample data set to be added.
Preferably, the picture classification for obtaining sample data set to be added, specifically includes:
It corresponds to table according to default picture classification to classify to the data set in sample data set to be added, with picture classification
The data set concentrated to the sample data to be added is named;
Obtain the picture classification of the sample data set to be added.
Preferably, crawl picture to be selected corresponding with the picture classification, specifically includes:
Grab Internet picture parameter information, using the corresponding Internet picture of the parameter information for meeting preset condition as
Picture to be selected corresponding with the picture classification.
Preferably, after crawl picture to be selected corresponding with the picture classification, the number of the sample data set
According to adding method further include:
The location information that the picture to be selected is extracted from the parameter information of the picture to be selected, the location information is deposited
Enter in document.
Preferably, described that the fisrt feature weighted value of each picture to be selected is determined by default neural network model, and obtain
The sample data to be added concentrates the second feature weighted value of storage, specifically includes:
The location information is read, the corresponding picture to be selected of the location information is obtained;
Each picture to be selected is substituted into default neural network model respectively, to obtain the fisrt feature weight of each picture to be selected
Value;
Obtain the second feature weighted value that the sample data to be added concentrates storage.
Preferably, described that each picture to be selected is substituted into default neural network model respectively, to obtain the of each picture to be selected
One feature weight value, specifically includes:
Each picture to be selected is traversed, using the picture traversed as current image;
Convolutional layer by presetting neural network model extracts the characteristic value of the current image, and the characteristic value is drawn
Characteristic pattern is made;
The main feature value in the characteristic pattern is extracted by the pond layer of the default neural network model;
Final characteristic value is calculated by the full connection layer functions of the default neural network model according to main feature value, and
Using the final characteristic value as the fisrt feature weighted value of the current image.
Preferably, it is described according to the fisrt feature weighted value and the second feature weighted value from the picture to be selected
Picture to be added is chosen, is specifically included:
The fisrt feature weighted value and the second feature weighted value are matched;
The feature weight value to match with the second feature weighted value is removed from the fisrt feature weighted value, will be remained
The corresponding picture to be selected of remaining fisrt feature weighted value is as picture to be added.
In addition, to achieve the above object, the present invention also proposes a kind of data adding set of sample data set, described device
Include:
Module is obtained, for obtaining the picture classification of sample data set to be added;
Handling module, for grabbing picture to be selected corresponding with the picture classification;
Determining module for determining the fisrt feature weighted value of each picture to be selected by default neural network model, and obtains
The sample data to be added is taken to concentrate the second feature weighted value of storage;
Choose module, for according to the fisrt feature weighted value and the second feature weighted value from the picture to be selected
It is middle to choose picture to be added;
Adding module, for the picture to be added to be added to the sample data set to be added.
In addition, to achieve the above object, the present invention also proposes a kind of terminal device, the terminal device include: memory,
Processor and the data addition program for being stored in the sample data set that can be run on the memory and on the processor,
The data addition program of the sample data set is arranged for carrying out the data adding method of sample data set as described above
Step.
In addition, to achieve the above object, the present invention also proposes that a kind of storage medium, the storage medium are computer storage
Medium is stored with the data addition program of sample data set, the data of the sample data set in the computer storage medium
The step of addition program realizes the data adding method of the sample data set when being executed by processor.
The invention discloses data adding method, device, terminal device and the storage medium of a kind of sample data set, packets
It includes: obtaining the picture classification of sample data set to be added;Grab picture to be selected corresponding with the picture classification;By default
Neural network model determines the fisrt feature weighted value of each picture to be selected, and obtains the sample data to be added and concentrate storage
Second feature weighted value;It is chosen from the picture to be selected according to the fisrt feature weighted value and the second feature weighted value
Picture to be added;The picture to be added is added to the sample data set to be added, the present invention by data constantly more
Newly, neural network is allow to extract the more features of same class picture from more pictures, to reach to data set
Addition and effect of optimization.
Detailed description of the invention
Fig. 1 is the structural schematic diagram of the terminal device for the hardware running environment that the embodiment of the present invention is related to;
Fig. 2 is the flow diagram of the data adding method first embodiment of sample of the present invention data set;
Fig. 3 is the flow diagram of the data adding method second embodiment of sample of the present invention data set;
Fig. 4 is the functional block diagram of the data adding method first embodiment of sample of the present invention data set.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
It should be appreciated that described herein, specific examples are only used to explain the present invention, is not intended to limit the present invention.
Referring to Fig.1, Fig. 1 is the structural representation of the terminal device for the hardware running environment that the embodiment of the present invention is related to
Figure.
As shown in Figure 1, the terminal device may include: processor 1001, such as central processing unit (Central
Processing Unit, CPU), communication bus 1002, user interface 1003, network interface 1004, memory 1005.Wherein,
Communication bus 1002 is for realizing the connection communication between these components.User interface 1003 may include display screen
(Display), input unit such as keyboard (Keyboard), optional user interface 1003 can also include that the wired of standard connects
Mouth, wireless interface.Network interface 1004 optionally may include standard wireline interface and wireless interface (such as Wireless Fidelity
(WIreless-FIdelity, WI-FI) interface).Memory 1005 can be the random access memory (Random of high speed
Access Memory, RAM) memory, be also possible to stable nonvolatile memory (Non-Volatile Memory,
), such as magnetic disk storage NVM.Memory 1005 optionally can also be the storage device independently of aforementioned processor 1001.
It will be understood by those skilled in the art that structure shown in Fig. 1 does not constitute the restriction to terminal device, in reality
Terminal device may include perhaps combining certain components or different components than illustrating more or fewer components in
Arrangement.
As shown in Figure 1, as may include operating system, network communication mould in a kind of memory 1005 of storage medium
The data of block, Subscriber Interface Module SIM and sample data set add program.
In terminal device shown in Fig. 1, network interface 1004 is mainly used for establishing terminal device and storage sample data
The communication connection of the server of all data needed for the data adding method system of collection;User interface 1003 be mainly used for
User carries out data interaction;Processor 1001, memory 1005 in the data adding method equipment of sample of the present invention data set
It can be set in the data-addition of sample data set, the data-addition of the sample data set passes through processor
The data addition program of the sample data set stored in 1001 calling memories 1005, and execute the present invention and the sample provided is provided
The data adding method of data set.
The embodiment of the invention provides a kind of data adding methods of sample data set, are sample of the present invention referring to Fig. 2, Fig. 2
The flow diagram of the data adding method first embodiment of notebook data collection.
In the present embodiment, the data adding method of the sample data set the following steps are included:
S10: the picture classification of sample data set to be added is obtained.
It should be noted that device end can pass through crawler before the picture classification for obtaining sample data set to be added
Engine obtains picture from internet, and the picture that will acquire is respectively put into training set file and test set text according to classification
In part folder, the training set file and test set file that the present invention is said all are data sets, are trained for the later period, work as picture
When having been placed in data set, device end can classify to picture according to the specific attribute of picture, and same class picture is placed on
In the same data set, and data set is named according to the classification of picture, device end of the invention can be obtained and ordered
The sample data set of name.
It should be understood that the crawler engine that the present invention is said is realized by web crawlers technology, web crawlers (and claimed
For webpage spider, network robot, more frequent is referred to as webpage follower), be it is a kind of according to certain rules, automatically grab
Take the program or script of web message.The rarely needed name of other there are also ant, automatic indexing, simulation program or
Person worm.
It should be understood that web crawlers is the program for automatically extracting webpage, it is search engine above and below WWW
Support grid page is the important composition of search engine.Uniform resource locator of traditional crawler from one or several Initial pages
(Uniform Resource Locator, URL) starts, and obtains the URL on Initial page, during grabbing webpage, no
It is disconnected to extract new URL from current page and be put into queue, certain stop condition until meeting system.The workflow of focused crawler
Journey is complex, needs to link according to certain web page analysis algorithm filtering is unrelated with theme, retains useful link and incite somebody to action
It such as is put at URL queue to be captured.Then, it will select next step to be grabbed according to certain search strategy from queue
Webpage URL, and repeat the above process, stopping when reaching a certain condition of system.In addition, all webpages by crawler capturing
It will be stored by system, certain analysis, filtering be carried out, and establish index, so as to inquiry and retrieval later;Focusing is climbed
For worm, the obtained analysis result of this process is also possible to provide feedback and guidance to later crawl process.
S20: crawl picture to be selected corresponding with the picture classification.
It should be noted that the present invention grabs the parameter information of Internet picture, the parameter information of preset condition will be met
Corresponding Internet picture is as picture to be selected corresponding with the picture classification.
It should be understood that capturing pictures compared with the prior art, do not obtain any information of picture, the present invention passes through
The parameter information of capturing pictures classifies to picture by the parameter information of picture, and is obtained by pre-set program
The picture to be selected for taking device end to need, the present invention, being capable of more efficient acquisition equipment ends by the parameter information of capturing pictures
Hold the picture to be selected needed.
S30: determining the fisrt feature weighted value of each picture to be selected by presetting neural network model, and obtains described wait add
It is loaded the centrally stored second feature weighted value of notebook data.
In the concrete realization, the location information is read, the corresponding picture to be selected of the location information is obtained, it will be each to be selected
Picture substitutes into default neural network model respectively, to obtain the fisrt feature weighted value of each picture to be selected, obtains described to be added
Sample data concentrates the second feature weighted value of storage.
It should be understood that the location information of picture to be selected is specially stored in a text document, when device end needs
When wanting picture, will from text document reading position information, and obtain the corresponding picture to be chosen of the location information.
It should be understood that each picture to be selected to be substituted into default neural network model respectively, to obtain each picture to be selected
Fisrt feature weighted value comprises the concrete steps that the convolutional layer by presetting neural network model extracts the feature of the current image
Value, and the characteristic value is depicted as characteristic pattern, the characteristic pattern is extracted by the pond layer of the default neural network model
In main feature value, calculated according to main feature value by the full connection layer functions of the default neural network model final special
Value indicative, and using the final characteristic value as the fisrt feature weighted value of the current image.
It should be understood that sample data concentration to be added may be stored with second feature weighted value, it is also possible to not deposit
Second feature weighted value is stored up, if device end is run for the first time, capturing pictures then add sample data and concentrate without storage the
Two feature weight values, after obtaining fisrt feature weighted value, device end can be weighed fisrt feature weighted value as second feature
Weight values are stored.
S40: chosen from the picture to be selected according to the fisrt feature weighted value and the second feature weighted value to
Add picture.
In the concrete realization, the fisrt feature weighted value and the second feature weighted value are matched, from described
The feature weight value to match with the second feature weighted value is removed in fisrt feature weighted value, and remaining fisrt feature is weighed
The corresponding picture to be selected of weight values is as picture to be added.
It should be understood that the present invention is by the fisrt feature weighted value and the second feature weighted value to be added
Sample data set carries out duplicate removal, when there are when duplicate data, pass through the comparison fisrt feature for sample data to be added concentration
Weighted value and the second feature weighted value, device end can remove the fisrt feature weighted value and the second feature weight
Identical feature weight value in value achievees the purpose that data update using different feature weight value as picture to be added.
S50: the picture to be added is added to the sample data set to be added.
In the concrete realization, the fisrt feature weighted value and the second feature weighted value are matched, from described
The feature weight value to match with the second feature weighted value is removed in fisrt feature weighted value, and remaining fisrt feature is weighed
Then picture to be added is added to the sample data set to be added as picture to be added by the corresponding picture to be selected of weight values,
Data update is carried out to the sample data set to be added.
The picture classification that the present embodiment passes through acquisition sample data set to be added;It grabs corresponding with the picture classification
Picture to be selected;The fisrt feature weighted value of each picture to be selected is determined by default neural network model, and is obtained described to be added
Sample data concentrates the second feature weighted value of storage;According to the fisrt feature weighted value and the second feature weighted value from
Picture to be added is chosen in the picture to be selected;The picture to be added is added to the sample data set to be added, this hair
It is bright so that neural network is extracted the more features of same class picture from more pictures by constantly updating to data,
To reach addition and the effect of optimization to data set.
Further, as shown in figure 3, proposing the data adding method of sample of the present invention data set based on first embodiment
Second embodiment.
In order to make it easy to understand, being specifically described below in conjunction with Fig. 3:
In step S20': the location information of the picture to be selected is extracted from the parameter information of the picture to be selected, it will
In the location information deposit document.
It should be understood that being stored with the location information of picture storage location, device end in the parameter information of picture to be selected
After reading a picture, the location information of the picture to be selected can be stored in a text document, equipment is whole after convenience
It takes at end.
It should be understood that the prior art for the selection of picture is directly used after grabbing, used picture
Can abandon, device end when needing image data, and can capturing pictures again, thus cause the waste of resource.This hair
In bright, device end can read the location information that picture storage location is stored in the parameter information of picture to be selected, and will be described
The location information of picture to be selected stores in a text document, facilitates the calling of follow-up equipment terminal, is greatly saved resource.
In addition, the embodiment of the present invention also proposes a kind of data adding set of sample data set.As shown in figure 4, the sample
The data adding set of data set includes: to obtain module 10, handling module 20, determining module 30, choose module 40, adding module
50。
Wherein, module 10 is obtained, for obtaining the picture classification of sample data set to be added;
Handling module 20, for grabbing picture to be selected corresponding with the picture classification;
Determining module 30, for determining the fisrt feature weighted value of each picture to be selected by presetting neural network model, and
Obtain the second feature weighted value that the sample data to be added concentrates storage;
Choose module 40, for according to the fisrt feature weighted value and the second feature weighted value from the figure to be selected
Picture to be added is chosen in piece;
Adding module 50, for the picture to be added to be added to the sample data set to be added.
By foregoing description it is not difficult to find that the picture classification that the present embodiment passes through acquisition sample data set to be added;Crawl
Picture to be selected corresponding with the picture classification;The fisrt feature power of each picture to be selected is determined by presetting neural network model
Weight values, and obtain the second feature weighted value that the sample data to be added concentrates storage;According to the fisrt feature weighted value
Picture to be added is chosen from the picture to be selected with the second feature weighted value;The picture to be added is added to described
Sample data set to be added, the present invention extract neural network from more pictures by constantly updating to data
The more features of same class picture, to reach the addition to data set and effect of optimization.
In addition, it should be noted that, the apparatus embodiments described above are merely exemplary, not to of the invention
Protection scope, which is constituted, to be limited, and in practical applications, those skilled in the art can select portion therein according to the actual needs
Point or whole module achieve the purpose of the solution of this embodiment, herein with no restrictions.
In addition, the not technical detail of detailed description in the present embodiment, reference can be made to provided by any embodiment of the invention
The data adding method of sample data set, details are not described herein again.
In addition, the embodiment of the present invention also proposes a kind of storage medium, the storage medium is computer storage medium, described
The data addition program of sample data set is stored in computer storage medium, the data of the sample data set add program quilt
Following operation is realized when processor executes:
Obtain the picture classification of sample data set to be added;
Grab picture to be selected corresponding with the picture classification;
The fisrt feature weighted value of each picture to be selected is determined by default neural network model, and obtains the sample to be added
The centrally stored second feature weighted value of notebook data;
It is chosen from the picture to be selected according to the fisrt feature weighted value and the second feature weighted value to be added
Picture;
The picture to be added is added to the sample data set to be added.
Further, the picture classification for obtaining sample data set to be added, specifically includes:
It corresponds to table according to default picture classification to classify to the data set in sample data set to be added, with picture classification
The data set concentrated to the sample data to be added is named;
Obtain the picture classification of the sample data set to be added.
Further, crawl picture to be selected corresponding with the picture classification, specifically includes:
Grab Internet picture parameter information, using the corresponding Internet picture of the parameter information for meeting preset condition as
Picture to be selected corresponding with the picture classification.
Further, after crawl picture to be selected corresponding with the picture classification, the sample data set
Data adding method further include:
The location information that the picture to be selected is extracted from the parameter information of the picture to be selected, the location information is deposited
Enter in document.
Further, described that the fisrt feature weighted value of each picture to be selected is determined by default neural network model, and obtain
It takes the sample data to be added to concentrate the second feature weighted value of storage, specifically includes:
The location information is read, the corresponding picture to be selected of the location information is obtained;
Each picture to be selected is substituted into default neural network model respectively, to obtain the fisrt feature weight of each picture to be selected
Value;
Obtain the second feature weighted value that the sample data to be added concentrates storage.
Further, described that each picture to be selected is substituted into default neural network model respectively, to obtain each picture to be selected
Fisrt feature weighted value, specifically includes:
Each picture to be selected is traversed, using the picture traversed as current image;
Convolutional layer by presetting neural network model extracts the characteristic value of the current image, and the characteristic value is drawn
Characteristic pattern is made;
The main feature value in the characteristic pattern is extracted by the pond layer of the default neural network model;
Final characteristic value is calculated by the full connection layer functions of the default neural network model according to main feature value, and
Using the final characteristic value as the fisrt feature weighted value of the current image.
Further, it is described according to the fisrt feature weighted value and the second feature weighted value from the picture to be selected
It is middle to choose picture to be added, it specifically includes:
The fisrt feature weighted value and the second feature weighted value are matched;
The feature weight value to match with the second feature weighted value is removed from the fisrt feature weighted value, will be remained
The corresponding picture to be selected of remaining fisrt feature weighted value is as picture to be added.
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that the process, method, article or the system that include a series of elements not only include those elements, and
And further include other elements that are not explicitly listed, or further include for this process, method, article or system institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do
There is also other identical elements in the process, method of element, article or system.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but in many cases
The former is more preferably embodiment.Based on this understanding, technical solution of the present invention substantially in other words does the prior art
The part contributed out can be embodied in the form of software products, which is stored in a storage medium
In (such as read-only memory/random access memory, magnetic disk, CD), including some instructions are used so that a terminal device (can
To be mobile phone, computer, server, air conditioner or the network equipment etc.) execute method described in each embodiment of the present invention.
The above is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (10)
1. a kind of data adding method of sample data set, which is characterized in that the data adding method packet of the sample data set
Include following steps:
Obtain the picture classification of sample data set to be added;
Grab picture to be selected corresponding with the picture classification;
The fisrt feature weighted value of each picture to be selected is determined by default neural network model, and obtains the sample number to be added
According to centrally stored second feature weighted value;
Picture to be added is chosen from the picture to be selected according to the fisrt feature weighted value and the second feature weighted value;
The picture to be added is added to the sample data set to be added.
2. the data adding method of sample data set as described in claim 1, which is characterized in that described to obtain sample to be added
The picture classification of data set, specifically includes:
It corresponds to table according to default picture classification to classify to the data set in sample data set to be added, to the sample to be added
The data set that notebook data is concentrated is named;
Obtain the picture classification of the sample data set to be added.
3. the data adding method of sample data set as described in claim 1, which is characterized in that the crawl and the picture
The corresponding picture to be selected of classification, specifically includes:
Grab Internet picture parameter information, using the corresponding Internet picture of the parameter information for meeting preset condition as with institute
State the corresponding picture to be selected of picture classification.
4. the data adding method of sample data set as claimed in claim 3, which is characterized in that the crawl and the picture
After the corresponding picture to be selected of classification, the data adding method of the sample data set further include:
The location information is stored in text by the location information that the picture to be selected is extracted from the parameter information of the picture to be selected
In shelves.
5. the data adding method of sample data set as described in claim 1, which is characterized in that described by presetting nerve net
Network model determines the fisrt feature weighted value of each picture to be selected, and obtains the second spy that the sample data to be added concentrates storage
Weighted value is levied, is specifically included:
The location information is read, the corresponding picture to be selected of the location information is obtained;
Each picture to be selected is substituted into default neural network model respectively, to obtain the fisrt feature weighted value of each picture to be selected;
Obtain the second feature weighted value that the sample data to be added concentrates storage.
6. the data adding method of sample data set as claimed in claim 5, which is characterized in that described by each picture to be selected point
Neural network model Dai Ru not be preset, to obtain the fisrt feature weighted value of each picture to be selected, is specifically included:
Each picture to be selected is traversed, using the picture traversed as current image;
Convolutional layer by presetting neural network model extracts the characteristic value of the current image, and the characteristic value is depicted as
Characteristic pattern;
The main feature value in the characteristic pattern is extracted by the pond layer of the default neural network model;
Final characteristic value is calculated by the full connection layer functions of the default neural network model according to main feature value, and by institute
State fisrt feature weighted value of the final characteristic value as the current image.
7. the data adding method of sample data set as described in claim 1, which is characterized in that described special according to described first
Sign weighted value and the second feature weighted value choose picture to be added from the picture to be selected, specifically include:
The fisrt feature weighted value and the second feature weighted value are matched;
The feature weight value to match with the second feature weighted value is removed from the fisrt feature weighted value, it will be remaining
The corresponding picture to be selected of fisrt feature weighted value is as picture to be added.
8. a kind of data adding set of sample data set, which is characterized in that described device includes:
Module is obtained, for obtaining the picture classification of sample data set to be added;
Handling module, for grabbing picture to be selected corresponding with the picture classification;
Determining module for determining the fisrt feature weighted value of each picture to be selected by default neural network model, and obtains institute
State the second feature weighted value that sample data to be added concentrates storage;
Module is chosen, for selecting from the picture to be selected according to the fisrt feature weighted value and the second feature weighted value
Take picture to be added;
Adding module, for the picture to be added to be added to the sample data set to be added.
9. a kind of terminal device, which is characterized in that the terminal device includes: memory, processor and is stored in described deposit
On reservoir and the data of the sample data set that can run on the processor addition program, the data of the sample data set add
The step of adding program to be arranged for carrying out the data adding method of sample data set as described in any one of claim 1 to 7.
10. a kind of storage medium, which is characterized in that the storage medium is computer storage medium, and the computer storage is situated between
The data addition program of sample data set is stored in matter, when the data addition program of the sample data set is executed by processor
The step of realizing the data adding method of sample data set as described in any one of claim 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910583761.7A CN110275974A (en) | 2019-06-28 | 2019-06-28 | Data adding method, device, terminal device and the storage medium of sample data set |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910583761.7A CN110275974A (en) | 2019-06-28 | 2019-06-28 | Data adding method, device, terminal device and the storage medium of sample data set |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110275974A true CN110275974A (en) | 2019-09-24 |
Family
ID=67964001
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910583761.7A Pending CN110275974A (en) | 2019-06-28 | 2019-06-28 | Data adding method, device, terminal device and the storage medium of sample data set |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110275974A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110990614A (en) * | 2019-11-08 | 2020-04-10 | 武汉东湖大数据交易中心股份有限公司 | Image self-learning method, device, equipment and medium based on engine big data |
CN111191119A (en) * | 2019-12-16 | 2020-05-22 | 绍兴市上虞区理工高等研究院 | Neural network-based scientific and technological achievement self-learning method and device |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107958272A (en) * | 2017-12-12 | 2018-04-24 | 北京旷视科技有限公司 | Image data set update method, device, system and computer-readable storage medium |
US9984312B2 (en) * | 2014-09-30 | 2018-05-29 | Samsung Electronics Co., Ltd. | Image registration device, image registration method, and ultrasonic diagnosis apparatus having image registration device |
CN108268533A (en) * | 2016-12-30 | 2018-07-10 | 南京烽火软件科技有限公司 | A kind of Image Feature Matching method for image retrieval |
CN109657694A (en) * | 2018-10-26 | 2019-04-19 | 平安科技(深圳)有限公司 | Picture automatic classification method, device and computer readable storage medium |
CN109918554A (en) * | 2019-02-13 | 2019-06-21 | 平安科技(深圳)有限公司 | Web data crawling method, device, system and computer readable storage medium |
-
2019
- 2019-06-28 CN CN201910583761.7A patent/CN110275974A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9984312B2 (en) * | 2014-09-30 | 2018-05-29 | Samsung Electronics Co., Ltd. | Image registration device, image registration method, and ultrasonic diagnosis apparatus having image registration device |
CN108268533A (en) * | 2016-12-30 | 2018-07-10 | 南京烽火软件科技有限公司 | A kind of Image Feature Matching method for image retrieval |
CN107958272A (en) * | 2017-12-12 | 2018-04-24 | 北京旷视科技有限公司 | Image data set update method, device, system and computer-readable storage medium |
CN109657694A (en) * | 2018-10-26 | 2019-04-19 | 平安科技(深圳)有限公司 | Picture automatic classification method, device and computer readable storage medium |
CN109918554A (en) * | 2019-02-13 | 2019-06-21 | 平安科技(深圳)有限公司 | Web data crawling method, device, system and computer readable storage medium |
Non-Patent Citations (1)
Title |
---|
孔令云: "《大数据关键技术与展望》", 28 February 2019, 四川大学出版社 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110990614A (en) * | 2019-11-08 | 2020-04-10 | 武汉东湖大数据交易中心股份有限公司 | Image self-learning method, device, equipment and medium based on engine big data |
CN111191119A (en) * | 2019-12-16 | 2020-05-22 | 绍兴市上虞区理工高等研究院 | Neural network-based scientific and technological achievement self-learning method and device |
CN111191119B (en) * | 2019-12-16 | 2023-12-12 | 绍兴市上虞区理工高等研究院 | Neural network-based scientific and technological achievement self-learning method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108733764B (en) | Advertisement filtering rule generation method based on machine learning and advertisement filtering system | |
CN104951539B (en) | Internet data center's harmful information monitoring system | |
CN105320740B (en) | The acquisition methods and acquisition system of wechat article and public platform | |
CN103430179B (en) | Add method, system and the computer-readable recording medium of new images and its relevant information in image data base | |
CN105243159A (en) | Visual script editor-based distributed web crawler system | |
CN105306495B (en) | user identification method and device | |
CN108574669B (en) | User behavior tree constructing method and device | |
CN103116638B (en) | Webpage screening method and device thereof | |
CN104915351A (en) | Picture sorting method and terminal | |
CN107092639A (en) | A kind of search engine system | |
CN102609412A (en) | RSS (Really Simple Syndication)-based multi-thread graphic information synchronization crawling control method and system | |
CN110413859A (en) | Webpage information search method, apparatus, computer equipment and storage medium | |
CN105302876A (en) | Regular expression based URL filtering method | |
CN109871770A (en) | Property ownership certificate recognition methods, device, equipment and storage medium | |
CN110275974A (en) | Data adding method, device, terminal device and the storage medium of sample data set | |
CN107330009A (en) | Descriptor disaggregated model creation method, creating device and storage medium | |
CN110490237A (en) | Data processing method, device, storage medium and electronic equipment | |
CN106682677A (en) | Advertising identification rule induction method, device and equipment | |
CN106649357A (en) | Data processing method and apparatus used for crawler program | |
CN106547803A (en) | The method and apparatus for crawling website incremental resource | |
CN109816004A (en) | Source of houses picture classification method, device, equipment and storage medium | |
CN110069686A (en) | User behavior analysis method, apparatus, computer installation and storage medium | |
CN107193870A (en) | The extracting method and system of web page contents | |
KR102257139B1 (en) | Method and apparatus for collecting information regarding dark web | |
CN103886033B (en) | Intelligent vertical searching device and method for safety industry chain |
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 |
Application publication date: 20190924 |
|
RJ01 | Rejection of invention patent application after publication |