CN108416003A - A kind of picture classification method and device, terminal, storage medium - Google Patents
A kind of picture classification method and device, terminal, storage medium Download PDFInfo
- Publication number
- CN108416003A CN108416003A CN201810164391.9A CN201810164391A CN108416003A CN 108416003 A CN108416003 A CN 108416003A CN 201810164391 A CN201810164391 A CN 201810164391A CN 108416003 A CN108416003 A CN 108416003A
- Authority
- CN
- China
- Prior art keywords
- target user
- picture
- label
- model
- current class
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
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
- G06F16/51—Indexing; Data structures therefor; Storage structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention discloses a kind of picture classification methods and device, terminal, storage medium, wherein, this method includes obtaining the label filtration model of target user, and label filtration model is obtained according to the advance training of history feature information of target user, the tag along sort for predicting target user；Using label filtration model, the current class tally set of target user is predicted according to the current characteristic information of target user；The picture of target user is identified, at least one picture classification collection corresponding with each tag along sort in current class tally set is obtained.The embodiment of the present invention may be implemented enrich existing picture classification method and meet user personalized classification demand effect.
Description
Technical field
The present embodiments relate to field of computer technology more particularly to a kind of picture classification method and device, terminal, deposit
Storage media.
Background technology
Currently, the picture number stored in various terminals is more and more, carrying out effective management to picture becomes to weigh very much
It wants.For example, in existing mobile terminal photograph album assorting process, realized using the labeling scheme of mobile terminal acquiescence mostly
The Classification Management of photograph album also needs to the manual customized label of user sometimes, and manual administration presses the photo of different dimensions tissue, and
One photo can only appear in a kind of classification of label.Based on the above classification schemes, it is primarily present following defect：
1) photograph album label dimension is limited：General only includes the simple label such as face or positioning, in face of different user
Ever-changing usage scenario, simple label extremely limit to the classification of photograph album.
2) scheme lacks autgmentability：Classification schemes are single, and classification extension cannot be carried out to current class scheme, such as：It presses
According to chronological classification, only divided according to the date, cannot be satisfied user according to the legal festivals and holidays, wedding anniversary or it is self-defined when
Between etc. special commemoration day carry out the classification extension of time dimension.
3) scheme lacks intelligence：Scheme cannot carry out personalized classification according to the characteristic of user itself, be unable to reach
" thousand people, thousand face ", is unable to the classification demand of accurate understanding user.
Therefore, existing picture classification scheme how is enriched, realizes that the personalized of picture is classified for different user demands,
It is still problem to be solved during pictures management.
Invention content
A kind of picture classification method of offer of the embodiment of the present invention and device, terminal, storage medium, it is abundant existing to realize
Picture classification method and meet user personalized classification demand effect.
In a first aspect, an embodiment of the present invention provides a kind of picture classification method, this method includes：
Obtain the label filtration model of target user, wherein the label filtration model is the history according to target user
Training obtains characteristic information in advance, the tag along sort for predicting target user；
Using the label filtration model, the current class of target user is predicted according to the current characteristic information of target user
Tally set；
The picture of target user is identified, is obtained and each tag along sort phase in the current class tally set
Corresponding at least one picture classification collection.
Second aspect, the embodiment of the present invention additionally provide a kind of picture classifier, which includes：
Label filtration model acquisition module, the label filtration model for obtaining target user, wherein the label filtration
Model is obtained according to the advance training of history feature information of target user, the tag along sort for predicting target user；
Current class tally set prediction module, for utilizing the label filtration model, according to the current spy of target user
Levy the current class tally set of information prediction target user；
Picture classification module is identified for the picture to target user, obtain in the current class tally set
The corresponding at least one picture classification collection of each tag along sort.
The third aspect, the embodiment of the present invention additionally provide a kind of terminal, including：
One or more processors；
Storage device, for storing one or more programs,
When one or more of programs are executed by one or more of processors so that one or more of processing
Device realizes the picture classification method as described in any embodiment of the present invention.
Fourth aspect, the embodiment of the present invention additionally provide a kind of computer readable storage medium, are stored thereon with computer
Program realizes the picture classification method as described in any embodiment of the present invention when the program is executed by processor.
The embodiment of the present invention is by obtaining the history feature information label filtration that training obtains in advance according to target user
Model predicts the current class tally set of target user according to the current characteristic information of target user, then to target user's
Picture is identified, and obtains at least one picture classification corresponding with each tag along sort in current class tally set
Collection.The embodiment of the present invention solves in the prior art that picture classification method is single, cannot be directed to user and realize personalized classification
Problem realizes abundant existing picture classification method and meets the effect of the personalized classification demand of user.
Description of the drawings
Fig. 1 is the flow chart for the picture classification method that the embodiment of the present invention one provides；
Fig. 2 is the flow chart of picture classification method provided by Embodiment 2 of the present invention；
Fig. 3 is the flow chart for the picture classification method that the embodiment of the present invention three provides；
Fig. 4 is the structural schematic diagram for the picture classifier that the embodiment of the present invention four provides；
Fig. 5 is a kind of structural schematic diagram for terminal that the embodiment of the present invention five provides.
Specific implementation mode
The present invention is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining the present invention rather than limitation of the invention.It also should be noted that in order to just
Only the parts related to the present invention are shown in description, attached drawing rather than entire infrastructure.
Embodiment one
Fig. 1 is the flow chart for the picture classification method that the embodiment of the present invention one provides, and the present embodiment is applicable to picture
The case where being classified, this method can be executed by picture classifier, which may be used the side of software and/or hardware
Formula is realized, and can be integrated in a kind of terminal, such as in the intellectual products such as computer and mobile terminal.As shown in Figure 1, this method has
Body includes：
S110, the label filtration model for obtaining target user, wherein label filtration model is the history according to target user
Training obtains characteristic information in advance, the tag along sort for predicting target user.
Target user can leave corresponding operation trace during using terminal, and terminal can correspond to operation trace
Data preserved, be then based on data analysis, the characteristic information of target user can be obtained.Characteristic information includes target
The interest of user and preference information etc. operate interest and preference that trace has reacted target user to a certain extent.For example,
For target user during using terminal, online search and browsing are relatively more about the picture of pet cat and the number of video,
Target user can be analyzed may be interested in pet cat.History feature information can be by periodically accumulating
The operation Trace Data of target user is analyzed to obtain, specific periodic time can according to the classification demand of target user into
Row setting.Based on obtained history feature information, the method that terminal can utilize machine learning trains to obtain label filtration mould
Type, and then predict the tag along sort of tendency when target user carries out picture classification, avoid showing for user's manual setting tag along sort
As improving the intelligence of picture classification.
S120, the label filtration model using acquisition predict target user's according to the current characteristic information of target user
Current class tally set.
The current characteristic information of target user is the direct reaction to the current state of target user, is believed based on current signature
Breath, the current class tally set of target user can be predicted using label filtration model.When current class tally set includes
Between class label, location category label, face class label, plant label, animal class label, weather class label, action class label, face
Color class label and more than at least two tag types combination etc., the label of each type can specifically be finely divided again, example
Such as, time class label includes legal festivals and holidays label, birthday label, commemoration day label and user-defined tag time mark
Label etc., location category label includes the location label of country variant and/or same country different cities.Illustratively, target user
Current characteristic information reaction target user it is interested in Tokyo oriental cherry, in the current class tally set of label filtration model prediction
There will be combination tag as the oriental cherry of Tokyo.
It is predicted by the current class tally set to target user, automatically generates picture classification label, can helped
Target user's efficiently picture in management terminal, realizes that the personalized photo for user's own characteristic is classified, also, current
Tag along sort concentration may include diversified combination tag, and combination tag increases the relevance between different labels, can
Corresponding to different usage scenarios, to realize the various dimensions extension to limited picture tag along sort in the prior art, solve
Single labeling scheme is spaced from cannot label be carried out the problem of flexible combination realizes classification in the prior art.
S130, the picture of target user is identified, is obtained and each tag along sort in current class tally set
Corresponding at least one picture classification collection.
After the current class tally set for determining target user, the picture of target user can be identified, pass through figure
The matching of piece characteristic information and the label information in tally set, picture is referred under corresponding label, picture classification collection is obtained.
For example, corresponding to combination tag --- the Tokyo oriental cherry of location category label and plant label, terminal can be to the institute of target user
There is picture to be identified, the picture classifications of the two characteristic informations of Tokyo and oriental cherry will be met simultaneously to Tokyo oriental cherry pictures
In.The picture of target user includes the picture for being stored in terminal local photograph album, also includes the picture in terminal cloud photograph album.When to end
When picture in the cloud photograph album of end is classified, need to realize the classification to picture in the photograph album of high in the clouds by network communication and transmission.
Wherein, after predicting the current class tally set of target user, terminal can be that each label is established centainly
Caching, the picture for storing the corresponding label temporarily will be stored with category images after completing whole picture classifications
The local storage space that data in caching are transferred to terminal respectively is stored, and the caching without category images is just released
It puts；Terminal can also carry out the foundation of corresponding picture classification collection when carrying out picture recognition, and the present embodiment does not limit this.
Picture classification in the photograph album of high in the clouds directly preserves picture classification collection beyond the clouds.Pass through picture automatic labeling and automatic point
Class gives bringing great convenience property of target user's search pictures, and improves search and search the timeliness of picture.
Based on the above technical solution, further, the training process of label filtration model includes：
Obtain operation behavior data of the target user in terminal；Optionally, operation behavior data include that target user exists
To the operation behavior data of picture, word and/or audio and video in terminal；
The behavioral parameters of target user are obtained according to the operation behavior data analysis of acquisition, wherein behavioral parameters are used for table
Levy the feature of target user；
The user of the behavioral parameters of acquisition and target user portrait are combined as target user's feature；
Using target user's feature as input, using to the tag along sort annotation results of target user's feature as exporting, it is sharp
It is trained to obtain label filtration model with the method for machine learning.
The operation behavior data of target user can be the operation behavior data to picture, word and/or audio and video, specifically
, for example, target user shoots picture and/or video, the picture of online browsing, word and/or audio and video in the recent period, or in public affairs
The operation behaviors data such as picture are shared or thumbed up to many platforms, these data are the equal of one to the interest of target user and preference
A statistics characterization.By carrying out image recognition, keyword extraction, semantic parsing or audio and video analysis etc. to operation behavior data,
The behavioral parameters of the feature of characterization target user are obtained, for example, behavioral parameters may include that the parameter that behavior represents (is such as shared, point
Praise or browse), the data semantic parameter (such as cat) they data type parameters (picture, word or the voice such as shared) and share
Deng.
A kind of virtual representations that user's portrait is networking products supplier to be formed according to different user's othernesses, such as hundred
User's portrait etc. is spent, the corresponding behavioral parameters of operation behavior data of target user and user's portrait are combined and are sieved as label
The training data of modeling type can improve the accuracy of model training.
Illustratively, target user frequently searches for or browses recently and will be incited somebody to action with the relevant picture of cat and article, terminal
Target user is characterized to the interested behavioral parameters of cat and user's portrait together as input, training obtains label filtration model,
The output result of label filtration model may be the tag along sort collection comprising label " cute pet " or label " cat ", work as users' mobile end
When be stored in photograph album with cat or the relevant picture of pet, it will be classified into the set of label " cute pet " or label " cat ".
The present embodiment technical solution is by obtaining the history feature information label that training obtains in advance according to target user
Screening model predicts the current class tally set of target user according to the current characteristic information of target user, then uses target
The picture at family is identified, and obtains at least one picture point corresponding with each tag along sort in current class tally set
Class set solves in the prior art the problem of picture classification method is single, cannot be directed to user's realization personalization classification, realizes
Effectively and accurately classified to the picture of target user, and enriches existing picture classification method and meet of user
The effect of propertyization classification demand so that picture classification result is bonded the feature and interest of target user；Also, current class label
Concentration may include diversified combination tag, and combination tag increases the relevance between different labels, can correspond to
Different usage scenarios, the single labeling scheme in the prior art that solves are spaced from cannot carry out label flexible
The problem of classification, is realized in combination, realizes the various dimensions extension to limited picture classification label in the prior art.
Embodiment two
Fig. 2 is the flow chart of picture classification method provided by Embodiment 2 of the present invention, and the present embodiment is in above-described embodiment
On the basis of further optimize.As shown in Fig. 2, this method specifically includes：
S210, the label filtration model for obtaining target user, wherein label filtration model is the history according to target user
Training obtains characteristic information in advance, the tag along sort for predicting target user.
S220, the label filtration model using acquisition predict target user's according to the current characteristic information of target user
Current class tally set.
The type of tag along sort, determination are corresponding with current class tally set extremely in S230, foundation current class tally set
A few image recognition model.
Since current class tally set includes diversified label, the corresponding picture of different type labels can lead to
Different recognizers is crossed to be identified, therefore, different type labels will correspond to different image recognition models, for example,
The corresponding picture of face class label can be identified by face recognition algorithms, and the corresponding picture of location category label can pass through
Ground point matching algorithm is identified, and the corresponding picture of face class label and the corresponding picture of location category label just correspond to different figures
As identification model.
Optionally, the training process of image recognition model includes：The history pictures cooperation that will be marked with tag along sort
It trains to obtain based on different classifications label using the method for machine learning using the tag along sort of mark as output for input
Image recognition model.
S240, the picture of target user is identified using at least one image recognition model, is obtained and current class
The corresponding at least one picture classification collection of each tag along sort in tally set.
The picture of target user is identified using different image recognition models, it is ensured that picture recognition it is accurate
Property, and then ensure the accuracy of picture classification.
Optionally, this method further includes periodically being updated to label filtration model and image recognition model by training.
Terminal is by regularly updating label filtration model and image recognition model, it is ensured that the figure in the present embodiment
The dynamic of piece sorting technique.With the accumulated change of the characteristic information of target user, the update that two models are adapted to is complete
It is kind, solve the problems, such as that picture classification method in the prior art cannot be changed and optimize into Mobile state, in addition, picture classification mistake
Cheng Zhong either predicts the current class tally set of target user, or using image recognition model to the picture of target user
It is identified, used model is updated model, ensure that the usage experience of user.
The technical solution of the present embodiment is by obtaining the history feature information mark that training obtains in advance according to target user
Screening model is signed, the current class tally set of target user is predicted according to the current characteristic information of target user, is then utilized not
The corresponding image recognition model of same type label is identified and classifies to the picture of target user, solves and schemes in the prior art
Piece sorting technique is single, cannot be directed to the problem of user realizes personalized classification, and the picture progress realized to target user is high
It imitates and accurately classifies, and enrich existing picture classification method and meet the effect of the personalized classification demand of user, this
Outside, perfect by regularly updating for label filtration model and image recognition model during picture classification, solve existing skill
The problem of picture classification method in art cannot be changed and optimize into Mobile state, ensure that the usage experience of user.
Embodiment three
Fig. 3 is the flow chart for the picture classification method that the embodiment of the present invention three provides, and the present embodiment is in above-described embodiment
On the basis of further optimize.As shown in figure 3, this method specifically includes：
S310, the label filtration model that target user is obtained from high in the clouds, wherein label filtration model are to be based on target user
High in the clouds data train to obtain and be updated beyond the clouds.
High in the clouds data based on target user realize the training of label filtration model beyond the clouds, can alleviate terminal training
Phenomena such as pressure of model time-histories sort run, the system interim card for avoiding terminal from occurring when more because of task process.Wherein, high in the clouds
Data are the information datas that can characterize target user's feature that server is collected from terminal, include the behavioral parameters of target user
It draws a portrait with user, after only being authorized by target user, high in the clouds can just utilize the high in the clouds data of target user, for example, working as
High in the clouds receives target user's mark of terminal transmission, is just considered to pass through mandate.After the training of label filtration model is completed in high in the clouds
It can be preserved, and timing updates label filtration model according to the variation of the high in the clouds data of target user.When pair for receiving terminal
After the acquisition request of label filtration model, current label filtration model will be issued to terminal.Meanwhile high in the clouds can be to working as
Whether preceding label filtration model, which updates, is judged, to ensure that be issued to terminal is newest label filtration model.
S320, the label filtration model using acquisition predict target user's according to the current characteristic information of target user
Current class tally set.
After terminal gets the label filtration model that the training that high in the clouds issues is completed, it is stored in local, according to target user
Current characteristic information, may be implemented the prediction of the current class tally set of target user, the wherein prediction can be online pre-
Survey can also be offline prediction, will not be limited by terminal current network state.
The type of tag along sort, determines and current class tally set pair from high in the clouds in S330, foundation current class tally set
At least one image recognition model answered, wherein image recognition model are to obtain simultaneously carrying out beyond the clouds more by training beyond the clouds
Newly.
Similar with label filtration model, the training of image recognition model, preservation and update are to carry out beyond the clouds, can be with
Avoid the memory headroom of occupied terminal.After the current class tally set of prediction determines, terminal sends out image recognition mould to high in the clouds
Image recognition model corresponding with concrete type label is just issued to end by the confirmation request of type, high in the clouds according to the confirmation request
End, terminal are stored in local after getting image recognition model.Equally, high in the clouds, can be to current image before issuing model
Whether identification model, which updates, is judged, to ensure that be issued to terminal is newest image recognition model.
S340, the picture of target user is identified using at least one image recognition model, is obtained and current class
The corresponding at least one picture classification collection of each tag along sort in tally set.
It should be noted that when the picture of terminal is stored in cloud photograph album, high in the clouds can be utilized directly by network communication
Label filtration model and image recognition model the picture in the cloud photograph album of target user is identified and is classified.
The technical solution of the present embodiment is believed by obtaining label filtration model from high in the clouds according to the current signature of target user
The current class tally set of breath prediction target user, then utilizes the figure of the image recognition model that is obtained from high in the clouds to target user
Piece is identified and classifies, and solves in the prior art that picture classification method is single, cannot be directed to user and realize personalized classification
The problem of, it realizes abundant existing picture classification method and meets the effect of the personalized classification demand of user, also, pass through
The training and update for completing label filtration model and image recognition model beyond the clouds, slow down the task pressure of terminal system operation
Power ensure that the fluency of terminal system operation, and model is issued to terminal from high in the clouds, may be implemented what picture was classified offline
Effect breaks away from constraint of the terminal internetwork connection condition to picture classification.
Example IV
Fig. 4 is the structural schematic diagram for the picture classifier that the embodiment of the present invention four provides, and the present embodiment is applicable to pair
The case where picture is classified.The picture classifier that the embodiment of the present invention is provided can perform any embodiment of the present invention and be carried
The picture classification method of confession has the corresponding function module of execution method and advantageous effect.As shown in figure 4, the device includes mark
Screening model acquisition module 410, current class tally set prediction module 420 and picture classification module 430 are signed, wherein：
Label filtration model acquisition module 410, the label filtration model for obtaining target user, wherein label filtration
Model is obtained according to the advance training of history feature information of target user, the tag along sort for predicting target user；
Current class tally set prediction module 420, for the label filtration model using acquisition, according to working as target user
The current class tally set of preceding characteristic information prediction target user；
Picture classification module 430 is identified for the picture to target user, obtain in current class tally set
The corresponding at least one picture classification collection of each tag along sort.
Further, picture classification module 430 includes：
Image recognition Models Sets determination unit, for the type according to tag along sort in current class tally set, determine with
The corresponding at least one image recognition model of current class tally set；
Picture classification unit is obtained for the picture of target user to be identified using at least one image recognition model
To at least one picture classification collection corresponding with each tag along sort in current class tally set.
Optionally, which further includes：
Label filtration model training module obtains the label filtration model for training；Wherein, the label filtration model
Training module includes：
Operation behavior data capture unit, for obtaining operation behavior data of the target user in terminal；
Behavioral parameters acquiring unit, the behavior for obtaining target user according to the operation behavior data analysis of acquisition are joined
Number, wherein behavioral parameters are used to characterize the feature of target user；
Target user's feature acquiring unit, for user's portrait of the behavioral parameters of acquisition and target user to be combined work
For target user's feature；
Model training unit is used for using target user's feature as input, will be to the tag along sort mark of target user's feature
Result is noted as output, trains to obtain label filtration model using the method for machine learning.
Optionally, operation behavior data capture unit is specifically used for：Target user is obtained in terminal to picture, word
And/or the operation behavior data of audio and video.
Optionally, which further includes：
Update module, for being periodically updated to label filtration model and image recognition model by training.
The embodiment of the present invention is by obtaining the history feature information label filtration that training obtains in advance according to target user
Model predicts the current class tally set of target user according to the current characteristic information of target user, then to target user's
Picture is identified, and obtains at least one picture classification corresponding with each tag along sort in current class tally set
Collection.The embodiment of the present invention solves in the prior art that picture classification method is single, cannot be directed to user and realize personalized classification
Problem realizes abundant existing picture classification method and meets the effect of the personalized classification demand of user.
Embodiment five
Fig. 5 is a kind of structural schematic diagram for terminal that the embodiment of the present invention five provides.Fig. 5 is shown suitable for being used for realizing this
The block diagram of the exemplary terminal 512 of invention embodiment.The terminal 512 that Fig. 5 is shown is only an example, should not be to the present invention
The function and use scope of embodiment bring any restrictions.
As shown in figure 5, terminal 512 is showed in the form of general purpose terminal.The component of terminal 512 can include but is not limited to：
One or more processor 516, storage device 528, connection different system component (including storage device 528 and processor
516) bus 518.
Bus 518 indicates one or more in a few class bus structures, including storage device bus or storage device control
Device processed, peripheral bus, graphics acceleration port, processor or total using the local of the arbitrary bus structures in a variety of bus structures
Line.For example, these architectures include but not limited to industry standard architecture (Industry Subversive
Alliance, ISA) bus, microchannel architecture (Micro Channel Architecture, MAC) bus is enhanced
Isa bus, Video Electronics Standards Association (Video Electronics Standards Association, VESA) local are total
Line and peripheral component interconnection (Peripheral Component Interconnect, PCI) bus.
Terminal 512 typically comprises a variety of computer system readable media.These media can be it is any can be by terminal
512 usable mediums accessed, including volatile and non-volatile media, moveable and immovable medium.
Storage device 528 may include the computer system readable media of form of volatile memory, such as arbitrary access
Memory (Random Access Memory, RAM) 530 and/or cache memory 532.Terminal 512 can be wrapped further
Include other removable/nonremovable, volatile/non-volatile computer system storage mediums.Only as an example, storage system
534 can be used for reading and writing immovable, non-volatile magnetic media (Fig. 5 do not show, commonly referred to as " hard disk drive ").Although
It is not shown, can be provided for the disc driver to moving non-volatile magnetic disk (such as " floppy disk ") read-write in Fig. 5, and
To moving anonvolatile optical disk, such as CD-ROM (Compact Disc Read-Only Memory, CD-ROM), number
Optic disk (Digital Video Disc-Read Only Memory, DVD-ROM) or other optical mediums) read-write CD drive
Dynamic device.In these cases, each driver can be connected by one or more data media interfaces with bus 518.It deposits
Storage device 528 may include at least one program product, which has one group of (for example, at least one) program module, this
A little program modules are configured to perform the function of various embodiments of the present invention.
Program/utility 540 with one group of (at least one) program module 542 can be stored in such as storage dress
In setting 528, such program module 542 includes but not limited to operating system, one or more application program, other program moulds
Block and program data may include the realization of network environment in each or certain combination in these examples.Program module
542 usually execute function and/or method in embodiment described in the invention.
Terminal 512 can also be logical with one or more external equipments 514 (such as keyboard, sensing equipment, display 524 etc.)
Letter, can also be enabled a user to one or more equipment interact with the terminal 512 communicate, and/or with make the terminal 512
Any equipment (such as network interface card, modem etc.) communication that can be communicated with one or more of the other computing device.This
Kind communication can be carried out by input/output (I/O) interface 522.Also, terminal 512 can also by network adapter 520 with
One or more network (such as LAN (Local Area Network, LAN), wide area network (Wide Area Network,
WAN) and/or public network, for example, internet) communication.As shown in figure 5, network adapter 520 passes through bus 518 and terminal 512
Other modules communication.It should be understood that although not shown in the drawings, other hardware and/or software mould can be used in conjunction with terminal 512
Block, including but not limited to：Microcode, device driver, redundant processor, external disk drive array, disk array
(Redundant Arrays of Independent Disks, RAID) system, tape drive and data backup storage system
System etc..
Processor 516 is stored in the program in storage device 528 by operation, to perform various functions application and number
According to processing, such as realize the picture classification method that the embodiment of the present invention is provided, this method includes：
Obtain the label filtration model of target user, wherein the label filtration model is the history according to target user
Training obtains characteristic information in advance, the tag along sort for predicting target user；
Using the label filtration model, the current class of target user is predicted according to the current characteristic information of target user
Tally set；
The picture of target user is identified, is obtained and each tag along sort phase in the current class tally set
Corresponding at least one picture classification collection.
Embodiment six
The embodiment of the present invention six additionally provides a kind of computer readable storage medium, is stored thereon with computer program, should
Realize that the picture classification method provided such as the embodiment of the present invention, this method include when program is executed by processor：
Obtain the label filtration model of target user, wherein the label filtration model is the history according to target user
Training obtains characteristic information in advance, the tag along sort for predicting target user；
Using the label filtration model, the current class of target user is predicted according to the current characteristic information of target user
Tally set；
The picture of target user is identified, is obtained and each tag along sort phase in the current class tally set
Corresponding at least one picture classification collection.
The arbitrary of one or more computer-readable media may be used in the computer storage media of the embodiment of the present invention
Combination.Computer-readable medium can be computer-readable signal media or computer readable storage medium.It is computer-readable
Storage medium for example may be-but not limited to-the system of electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor, device or
Device, or the arbitrary above combination.The more specific example (non exhaustive list) of computer readable storage medium includes：Tool
There are one or the electrical connection of multiple conducting wires, portable computer diskette, hard disk, random access memory (RAM), read-only memory
(ROM), erasable programmable read only memory (Erasable Programmable Read Only Memory, EPROM, or
Flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light storage device, magnetic memory device or above-mentioned
Any appropriate combination.In this document, can be any include computer readable storage medium or tangible Jie of storage program
Matter, the program can be commanded the either device use or in connection of execution system, device.
Computer-readable signal media may include in a base band or as the data-signal that a carrier wave part is propagated,
Wherein carry computer-readable program code.Diversified forms may be used in the data-signal of this propagation, including but unlimited
In electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be that computer can
Any computer-readable medium other than storage medium is read, which can send, propagates or transmit and be used for
By instruction execution system, device either device use or program in connection.
The program code for including on computer-readable medium can transmit with any suitable medium, including --- but it is unlimited
In wireless, electric wire, optical cable, radio frequency (Radio Frequency, RF) etc. or above-mentioned any appropriate combination.
It can be write with one or more programming languages or combinations thereof for executing the computer that operates of the present invention
Program code, described program design language include object oriented program language-such as Java, Smalltalk, C++,
Further include conventional procedural programming language-such as " C " language or similar programming language.Program code can be with
It fully executes, partly execute on the user computer on the user computer, being executed as an independent software package, portion
Divide and partly executes or executed on remote computer or terminal completely on the remote computer on the user computer.It is relating to
And in the situation of remote computer, remote computer can pass through the network of any kind --- including LAN (LAN) or extensively
Domain net (WAN)-be connected to subscriber computer, or, it may be connected to outer computer (such as provided using Internet service
Quotient is connected by internet).
Note that above are only presently preferred embodiments of the present invention and institute's application technology principle.It will be appreciated by those skilled in the art that
The present invention is not limited to specific embodiments described here, can carry out for a person skilled in the art it is various it is apparent variation,
It readjusts and substitutes without departing from protection scope of the present invention.Therefore, although being carried out to the present invention by above example
It is described in further detail, but the present invention is not limited only to above example, without departing from the inventive concept, also
May include other more equivalent embodiments, and the scope of the present invention is determined by scope of the appended claims.
Claims (12)
1. a kind of picture classification method, which is characterized in that including：
Obtain the label filtration model of target user, wherein the label filtration model is the history feature according to target user
Training obtains information in advance, the tag along sort for predicting target user；
Using the label filtration model, the current class label of target user is predicted according to the current characteristic information of target user
Collection；
The picture of target user is identified, is obtained corresponding with each tag along sort in the current class tally set
At least one picture classification collection.
2. according to the method described in claim 1, it is characterized in that, the training process of the label filtration model includes：
Obtain operation behavior data of the target user in terminal；
The behavioral parameters of target user are obtained according to the operation behavior data analysis, wherein the behavioral parameters are for characterizing
The feature of target user；
The user of the behavioral parameters and target user portrait are combined as target user's feature；
Using target user's feature as input, using to the tag along sort annotation results of target user's feature as defeated
Go out, trains to obtain the label filtration model using the method for machine learning.
3. according to the method described in claim 2, it is characterized in that, the operation behavior data include target user in terminal
To the operation behavior data of picture, word and/or audio and video.
4. according to the method described in claim 1, it is characterized in that, the picture to target user is identified, obtain with
The corresponding at least one picture classification collection of each tag along sort in the current class tally set, including：
According to the type of tag along sort in the current class tally set, determination is corresponding at least with the current class tally set
One image recognition model；
The picture of target user is identified using at least one image recognition model, is obtained and the current class mark
Sign the corresponding at least one picture classification collection of each tag along sort concentrated.
5. according to the method described in claim 4, it is characterized in that, the method further includes：
Periodically the label filtration model and image recognition model are updated by training.
6. a kind of picture classifier, which is characterized in that including：
Label filtration model acquisition module, the label filtration model for obtaining target user, wherein the label filtration model
It is to be obtained according to the advance training of history feature information of target user, the tag along sort for predicting target user；
Current class tally set prediction module is believed for utilizing the label filtration model according to the current signature of target user
The current class tally set of breath prediction target user；
Picture classification module is identified for the picture to target user, obtain with it is every in the current class tally set
The corresponding at least one picture classification collection of one tag along sort.
7. device according to claim 6, which is characterized in that described device further includes：
Label filtration model training module obtains the label filtration model for training；Wherein, the label filtration model instruction
Practicing module includes：
Operation behavior data capture unit, for obtaining operation behavior data of the target user in terminal；
Behavioral parameters acquiring unit, for obtaining the behavioral parameters of target user according to the operation behavior data analysis, wherein
The behavioral parameters are used to characterize the feature of target user；
Target user's feature acquiring unit, for being combined user's portrait of the behavioral parameters and target user as target
User characteristics；
Model training unit is used for using target user's feature as input, will be to the contingency table of target user's feature
Annotation results are signed as output, train to obtain the label filtration model using the method for machine learning.
8. device according to claim 7, which is characterized in that the operation behavior data capture unit is specifically used for：It obtains
Take operation behavior data of the target user to picture, word and/or audio and video in terminal.
9. device according to claim 6, which is characterized in that the picture classification module includes：
Image recognition Models Sets determination unit, for type according to tag along sort in the current class tally set, determine with
The corresponding at least one image recognition model of the current class tally set；
Picture classification unit is obtained for the picture of target user to be identified using at least one image recognition model
To at least one picture classification collection corresponding with each tag along sort in the current class tally set.
10. device according to claim 9, which is characterized in that described device further includes：
Update module, for being periodically updated to the label filtration model and image recognition model by training.
11. a kind of terminal, which is characterized in that including：
One or more processors；
Storage device, for storing one or more programs,
When one or more of programs are executed by one or more of processors so that one or more of processors are real
The now picture classification method as described in any in Claims 1 to 5.
12. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is by processor
The picture classification method as described in any in Claims 1 to 5 is realized when execution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810164391.9A CN108416003B (en) | 2018-02-27 | 2018-02-27 | Picture classification method and device, terminal and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810164391.9A CN108416003B (en) | 2018-02-27 | 2018-02-27 | Picture classification method and device, terminal and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108416003A true CN108416003A (en) | 2018-08-17 |
| CN108416003B CN108416003B (en) | 2021-12-21 |
Family
ID=63129115
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810164391.9A Active CN108416003B (en) | 2018-02-27 | 2018-02-27 | Picture classification method and device, terminal and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108416003B (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109446783A (en) * | 2018-11-16 | 2019-03-08 | 济南浪潮高新科技投资发展有限公司 | A kind of efficient sample collection method and system of image recognition based on machine crowdsourcing |
| CN109597912A (en) * | 2018-12-05 | 2019-04-09 | 上海碳蓝网络科技有限公司 | Method for handling picture |
| CN110110750A (en) * | 2019-03-29 | 2019-08-09 | 广州思德医疗科技有限公司 | A kind of classification method and device of original image |
| CN110390033A (en) * | 2019-07-25 | 2019-10-29 | 腾讯科技（深圳）有限公司 | Training method, device, electronic equipment and the storage medium of image classification model |
| CN110874420A (en) * | 2019-11-09 | 2020-03-10 | 上海光数信息科技有限公司 | Picture information processing system and method |
| WO2020063042A1 (en) * | 2018-09-26 | 2020-04-02 | Oppo广东移动通信有限公司 | Picture classification method and apparatus, and computer-readable storage medium and electronic device |
| CN111027557A (en) * | 2019-03-11 | 2020-04-17 | 广东小天才科技有限公司 | Subject identification method based on subject image and electronic equipment |
| CN111143568A (en) * | 2019-12-31 | 2020-05-12 | 郑州工程技术学院 | Method, device and equipment for buffering during paper classification and storage medium |
| CN111274427A (en) * | 2020-01-08 | 2020-06-12 | 上海擎感智能科技有限公司 | Picture processing method and device and computer storage medium |
| CN111399778A (en) * | 2020-03-17 | 2020-07-10 | 西安石油大学 | Picture file cleaning method and device |
| CN111723226A (en) * | 2020-06-14 | 2020-09-29 | 黄雨勤 | Information management method based on big data and Internet and artificial intelligence cloud server |
| CN111966850A (en) * | 2020-07-21 | 2020-11-20 | 珠海格力电器股份有限公司 | Picture screening method and device, electronic equipment and storage medium |
| CN112288793A (en) * | 2020-11-06 | 2021-01-29 | 洛阳语音云创新研究院 | Livestock individual backfat detection method and device, electronic equipment and storage medium |
| CN112580681A (en) * | 2019-09-30 | 2021-03-30 | 北京星选科技有限公司 | User classification method and device, electronic equipment and readable storage medium |
| CN113496236A (en) * | 2020-03-20 | 2021-10-12 | 北京沃东天骏信息技术有限公司 | User tag information determination method, device, equipment and storage medium |
| CN113936206A (en) * | 2020-07-08 | 2022-01-14 | 华为技术有限公司 | Method, medium, and system for classifying image on terminal device |
| CN114047996A (en) * | 2021-11-25 | 2022-02-15 | 维沃移动通信有限公司 | Wallpaper setting method, wallpaper setting device and electronic equipment |
| CN115019218A (en) * | 2022-08-08 | 2022-09-06 | 阿里巴巴（中国）有限公司 | Image processing method and processor |
| CN116127108A (en) * | 2023-02-20 | 2023-05-16 | 广东德融汇科技有限公司 | Album generation method and device based on image recognition and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103995889A (en) * | 2014-06-03 | 2014-08-20 | 广东欧珀移动通信有限公司 | Method and device for classifying pictures |
| US20150161170A1 (en) * | 2009-07-17 | 2015-06-11 | Google Inc. | Image classification |
| CN105760448A (en) * | 2016-02-03 | 2016-07-13 | 北京金山安全软件有限公司 | Picture processing method and device and electronic equipment |
| CN106055617A (en) * | 2016-05-26 | 2016-10-26 | 乐视控股（北京）有限公司 | Data pushing method and device |
| CN106445995A (en) * | 2016-07-18 | 2017-02-22 | 腾讯科技（深圳）有限公司 | Picture classification method and apparatus |
| CN106909629A (en) * | 2017-01-26 | 2017-06-30 | 武汉奇米网络科技有限公司 | A kind of personalized recommendation cheats the method and system of position label |
| CN107341245A (en) * | 2017-07-06 | 2017-11-10 | 广州优视网络科技有限公司 | Data processing method, device and server |
-
2018
- 2018-02-27 CN CN201810164391.9A patent/CN108416003B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150161170A1 (en) * | 2009-07-17 | 2015-06-11 | Google Inc. | Image classification |
| CN103995889A (en) * | 2014-06-03 | 2014-08-20 | 广东欧珀移动通信有限公司 | Method and device for classifying pictures |
| CN105760448A (en) * | 2016-02-03 | 2016-07-13 | 北京金山安全软件有限公司 | Picture processing method and device and electronic equipment |
| CN106055617A (en) * | 2016-05-26 | 2016-10-26 | 乐视控股（北京）有限公司 | Data pushing method and device |
| CN106445995A (en) * | 2016-07-18 | 2017-02-22 | 腾讯科技（深圳）有限公司 | Picture classification method and apparatus |
| CN106909629A (en) * | 2017-01-26 | 2017-06-30 | 武汉奇米网络科技有限公司 | A kind of personalized recommendation cheats the method and system of position label |
| CN107341245A (en) * | 2017-07-06 | 2017-11-10 | 广州优视网络科技有限公司 | Data processing method, device and server |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020063042A1 (en) * | 2018-09-26 | 2020-04-02 | Oppo广东移动通信有限公司 | Picture classification method and apparatus, and computer-readable storage medium and electronic device |
| CN109446783A (en) * | 2018-11-16 | 2019-03-08 | 济南浪潮高新科技投资发展有限公司 | A kind of efficient sample collection method and system of image recognition based on machine crowdsourcing |
| CN109597912A (en) * | 2018-12-05 | 2019-04-09 | 上海碳蓝网络科技有限公司 | Method for handling picture |
| CN111027557A (en) * | 2019-03-11 | 2020-04-17 | 广东小天才科技有限公司 | Subject identification method based on subject image and electronic equipment |
| CN111027557B (en) * | 2019-03-11 | 2024-03-19 | 广东小天才科技有限公司 | Subject identification method based on subject image and electronic equipment |
| CN110110750A (en) * | 2019-03-29 | 2019-08-09 | 广州思德医疗科技有限公司 | A kind of classification method and device of original image |
| CN110390033A (en) * | 2019-07-25 | 2019-10-29 | 腾讯科技（深圳）有限公司 | Training method, device, electronic equipment and the storage medium of image classification model |
| CN110390033B (en) * | 2019-07-25 | 2023-04-21 | 腾讯科技（深圳）有限公司 | Training method and device for image classification model, electronic equipment and storage medium |
| CN112580681A (en) * | 2019-09-30 | 2021-03-30 | 北京星选科技有限公司 | User classification method and device, electronic equipment and readable storage medium |
| CN110874420A (en) * | 2019-11-09 | 2020-03-10 | 上海光数信息科技有限公司 | Picture information processing system and method |
| CN111143568A (en) * | 2019-12-31 | 2020-05-12 | 郑州工程技术学院 | Method, device and equipment for buffering during paper classification and storage medium |
| CN111274427A (en) * | 2020-01-08 | 2020-06-12 | 上海擎感智能科技有限公司 | Picture processing method and device and computer storage medium |
| CN111399778A (en) * | 2020-03-17 | 2020-07-10 | 西安石油大学 | Picture file cleaning method and device |
| CN113496236B (en) * | 2020-03-20 | 2024-05-24 | 北京沃东天骏信息技术有限公司 | User tag information determining method, device, equipment and storage medium |
| CN113496236A (en) * | 2020-03-20 | 2021-10-12 | 北京沃东天骏信息技术有限公司 | User tag information determination method, device, equipment and storage medium |
| CN111723226A (en) * | 2020-06-14 | 2020-09-29 | 黄雨勤 | Information management method based on big data and Internet and artificial intelligence cloud server |
| CN113936206A (en) * | 2020-07-08 | 2022-01-14 | 华为技术有限公司 | Method, medium, and system for classifying image on terminal device |
| CN111966850A (en) * | 2020-07-21 | 2020-11-20 | 珠海格力电器股份有限公司 | Picture screening method and device, electronic equipment and storage medium |
| CN112288793A (en) * | 2020-11-06 | 2021-01-29 | 洛阳语音云创新研究院 | Livestock individual backfat detection method and device, electronic equipment and storage medium |
| CN114047996A (en) * | 2021-11-25 | 2022-02-15 | 维沃移动通信有限公司 | Wallpaper setting method, wallpaper setting device and electronic equipment |
| CN115019218A (en) * | 2022-08-08 | 2022-09-06 | 阿里巴巴（中国）有限公司 | Image processing method and processor |
| CN116127108A (en) * | 2023-02-20 | 2023-05-16 | 广东德融汇科技有限公司 | Album generation method and device based on image recognition and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108416003B (en) | 2021-12-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108416003A (en) | A kind of picture classification method and device, terminal, storage medium | |
| CN103959282B (en) | For the selective feedback of text recognition system | |
| CN110134931B (en) | Medium title generation method, medium title generation device, electronic equipment and readable medium | |
| CN109919244B (en) | Method and apparatus for generating a scene recognition model | |
| CN107251060A (en) | For the pre-training and/or transfer learning of sequence label device | |
| CN109947989B (en) | Method and apparatus for processing video | |
| US20210209289A1 (en) | Method and apparatus for generating customized content based on user intent | |
| CN108509569A (en) | Generation method, device, electronic equipment and the storage medium of enterprise's portrait | |
| CN108369806B (en) | Configurable generic language understanding model | |
| CN112215171B (en) | Target detection method, device, equipment and computer readable storage medium | |
| CN111126390A (en) | Correlation method and device for identifying identification pattern in media content | |
| US11462018B2 (en) | Representative image generation | |
| CN109165316A (en) | A kind of method for processing video frequency, video index method, device and terminal device | |
| CN113033682B (en) | Video classification method, device, readable medium and electronic equipment | |
| CN109389660A (en) | Image generating method and device | |
| CN116955591A (en) | Recommendation language generation method, related device and medium for content recommendation | |
| CN112070093B (en) | Method for generating image classification model, image classification method, device and equipment | |
| CN110442803A (en) | Data processing method, device, medium and the calculating equipment executed by calculating equipment | |
| CN109934194A (en) | Picture classification method, edge device, system and storage medium | |
| CN112685578A (en) | Multimedia information content providing method and device | |
| CN107291774A (en) | Error sample recognition methods and device | |
| CN111444335A (en) | Method and device for extracting central word | |
| CN110909506A (en) | Method and device for generating planar design material | |
| US20220189472A1 (en) | Recognition and restructuring of previously presented materials | |
| CN114840743B (en) | Model recommendation method and device, electronic equipment and readable storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |