CN108491866A - Porny identification method, electronic device and readable storage medium storing program for executing - Google Patents
Porny identification method, electronic device and readable storage medium storing program for executing Download PDFInfo
- Publication number
- CN108491866A CN108491866A CN201810184583.6A CN201810184583A CN108491866A CN 108491866 A CN108491866 A CN 108491866A CN 201810184583 A CN201810184583 A CN 201810184583A CN 108491866 A CN108491866 A CN 108491866A
- Authority
- CN
- China
- Prior art keywords
- picture
- porny
- identified
- probability value
- format
- 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
- 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- 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
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural 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
- G06N3/04—Architecture, e.g. interconnection topology
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Multimedia (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Algebra (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of porny identification method, electronic device and readable storage medium storing program for executing, this method to include：After receiving picture to be identified, the picture format of the picture to be identified is detected, and the picture to be identified is identified using picture classification model trained in advance, exports the picture to be identified and belong to the other probability value of one or more default picture categories；Wherein, the picture classification model is convolutional neural networks model, which advances with the other samples pictures of default picture category and be trained to obtain；According to the picture format of the picture to be identified and the probability value, judge whether the picture to be identified belongs to porny.The present invention can accurately and effectively judge whether picture to be identified belongs to porny；And it is not necessarily to artificial detection, porny identification can be carried out automatically, effectively improve detection efficiency.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of porny identification method, electronic devices and readable
Storage medium.
Background technology
Currently, for Large-Scale Interconnected net financial company, a large number of services picture can be related to, and meeting is possible in business picture
It is mingled with porny, makes a very bad impression, it is necessary to effectively identifies and reject.Traditional porny identification mode is by manually to business
Picture is checked that this artificial detection is of high cost, and than relatively time-consuming to filter out porny therein one by one, efficiency compared with
It is low.
Invention content
The purpose of the present invention is to provide a kind of porny identification method, electronic device and readable storage medium storing program for executing, it is intended to
Improve the efficiency of identification porny.
To achieve the above object, the present invention provides a kind of porny identification method, the porny identification method packet
It includes：
After receiving picture to be identified, the picture format of the picture to be identified is detected, and utilize picture trained in advance
The picture to be identified is identified in disaggregated model, exports the picture to be identified and belongs to one or more default picture classifications
Probability value；Wherein, the picture classification model is convolutional neural networks model, which is to advance with
What the other samples pictures of default picture category were trained；
According to the picture format of the picture to be identified and the probability value, judge whether the picture to be identified belongs to color
Feelings picture.
Preferably, the picture format includes black and white format and non-black and white format, and the default picture classification includes pornographic
Picture and non-porny；
The output picture to be identified belongs to the step of one or more default picture categories other probability value and specifically wraps
It includes：Export the probability value that the picture to be identified belongs to the probability value and non-porny of porny；
The picture format according to the picture to be identified and the probability value, judge whether the picture to be identified belongs to
It is specifically included in the step of porny：
Reduce the probability value that the picture that picture format is black and white format belongs to porny；
Judge whether the picture of the black and white format belongs to porny according to the probability value after reduction, and according to described
Probability value judges whether the picture of non-black and white format belongs to porny.
Preferably, the probability value according to after reduction judges whether the picture of the black and white format belongs to porny
Step specifically includes：
Probability value after the reduction is compared with the first predetermined threshold value, if it is described turn down after probability value be more than institute
Preset first threshold value is stated, then judges that the picture to be identified belongs to porny；
If the probability value after turning down is less than or equal to the preset first threshold value, judge that the picture to be identified is not belonging to
Porny.
Preferably, the picture format includes black and white format and non-black and white format, and the default picture classification includes pornographic
Picture, sexy picture and normal picture；
The output picture to be identified belongs to the step of one or more default picture categories other probability value and specifically wraps
It includes：Export probability value, the sexy probability value of picture and the probability of normal picture that the picture to be identified belongs to porny
Value；
The picture format according to the picture to be identified and the probability value, judge whether the picture to be identified belongs to
It is specifically included in the step of porny：
Reduce the probability value that the picture that picture format is black and white format belongs to porny；
Judge whether the picture of the black and white format belongs to porny according to the probability value after reduction, and according to described
Probability value judges whether the picture of non-black and white format belongs to porny.
Preferably, the probability value according to after reduction judges whether the picture of the black and white format belongs to porny
Step specifically includes：
Probability value after the reduction is compared with the second predetermined threshold value, if it is described turn down after probability value be more than institute
Default second threshold is stated, then judges that the picture to be identified belongs to porny；
If the probability value after turning down is less than or equal to the default second threshold, judge that the picture to be identified is not belonging to
Porny.
Preferably, the training step of the picture classification model is as follows：
A, porny, sexy picture and the normal picture of preset quantity are collected as samples pictures, and in each sample
Corresponding picture classification is marked on picture；
B, picture pretreatment is carried out to each samples pictures；
C, pretreated samples pictures are divided into the verification collection of the training set and the second ratio of the first ratio；
D, the training set training convolutional neural networks model is utilized；
E, using the accuracy rate of the convolutional neural networks model of the verification collection verification training, if accuracy rate is more than or waits
In default accuracy rate, then training terminates, and using trained convolutional neural networks model as picture classification model；If accuracy rate
Less than default accuracy rate, then samples pictures are collected again and repeat above-mentioned steps B, C, D, E.
Preferably, further comprising the steps of when the picture to be identified is dynamic picture：
The picture to be identified is cut into frame into plurality of pictures, detection cuts the picture format of each pictures of frame, and respectively
The plurality of pictures for cutting frame is identified using picture classification model trained in advance, each pictures that output cuts frame belong to each
A other probability value of default picture category；
According to the picture format of each pictures of described section of frame and the probability value, whether the picture to be identified is judged
Belong to porny.
Preferably, the picture format of each pictures of frame and the probability value are cut described in the basis, are waited for described in judgement
The step of whether identification picture belongs to porny specifically includes：
According to the picture format of each pictures of described section of frame and the probability value, judge whether deposited in the picture of section frame
In the picture for belonging to porny；
If in the presence of judging that the picture to be identified belongs to porny；Otherwise, judge that the picture to be identified is not belonging to
Porny；Or
According to the picture format of each pictures of described section of frame and the probability value, pornographic figure in the picture of section frame is judged
Whether piece probability value is more than the picture of preset value more than two；
If so, the judgement picture to be identified belongs to porny；Otherwise, judge that the picture to be identified is not belonging to color
Feelings picture.
In addition, to achieve the above object, the present invention also provides a kind of electronic device, the electronic device include memory,
Processor is stored with the porny identification systems that can be run on the processor, the porny on the memory
Identification systems realize following steps when being executed by the processor：
After receiving picture to be identified, the picture format of the picture to be identified is detected, and utilize picture trained in advance
The picture to be identified is identified in disaggregated model, exports the picture to be identified and belongs to one or more default picture classifications
Probability value；Wherein, the picture classification model is convolutional neural networks model, which is to advance with
What the other samples pictures of default picture category were trained；
According to the picture format of the picture to be identified and the probability value, judge whether the picture to be identified belongs to color
Feelings picture.
Further, to achieve the above object, the present invention also provides a kind of computer readable storage medium, the computers
Readable storage medium storing program for executing is stored with porny identification systems, and the porny identification systems can be held by least one processor
Row, so that at least one processor is executed such as the step of above-mentioned porny identification method.
Porny identification method, system and readable storage medium storing program for executing proposed by the present invention pass through picture point trained in advance
Class model is treated identification picture and is identified, and exports the picture to be identified and belongs to each other probability of default picture category；And it examines
Survey the picture format of the picture to be identified；Belong to the other probability of each default picture category and institute according to the picture to be identified
The picture format of picture to be identified is stated, and judges whether the picture to be identified belongs to porny based on preset rules.Due to
The present invention can recognize that picture to be identified belongs to the other probability of one or more default picture categories, and combine the figure of picture to be identified
Piece format carries out comprehensive identification, can more accurately and effectively judge whether the picture to be identified belongs to porny.And
And without being manually detected, porny identification can be carried out automatically, effectively improves detection efficiency.
Description of the drawings
Fig. 1 is the running environment schematic diagram of 10 preferred embodiment of porny identification systems of the present invention；
Fig. 2 is the flow diagram of one embodiment of porny identification method of the present invention.
Specific implementation mode
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that described herein, specific examples are only used to explain the present invention, not
For limiting the present invention.Based on the embodiments of the present invention, those of ordinary skill in the art are not before making creative work
The every other embodiment obtained is put, shall fall within the protection scope of the present invention.
It should be noted that the description for being related to " first ", " second " etc. in the present invention is used for description purposes only, and cannot
It is interpreted as indicating or implying its relative importance or implicitly indicates the quantity of indicated technical characteristic.Define as a result, " the
One ", the feature of " second " can explicitly or implicitly include at least one of the features.In addition, the skill between each embodiment
Art scheme can be combined with each other, but must can be implemented as basis with those of ordinary skill in the art, when technical solution
Will be understood that the combination of this technical solution is not present in conjunction with there is conflicting or cannot achieve when, also not the present invention claims
Protection domain within.
The present invention provides a kind of porny identification systems.Referring to Fig. 1, be porny identification systems 10 of the present invention compared with
The running environment schematic diagram of good embodiment.
In the present embodiment, the porny identification systems 10 are installed and are run in electronic device 1.The electronics fills
It sets 1 may include, but is not limited only to, memory 11, processor 12 and display 13.Fig. 1 illustrates only the electricity with component 11-13
Sub-device 1, it should be understood that being not required for implementing all components shown, the implementation that can be substituted is more or less
Component.
The memory 11 is the readable computer storage medium of at least one type, and the memory 11 is implemented at some
Can be the internal storage unit of the electronic device 1, such as the hard disk or memory of the electronic device 1 in example.The memory
11 can also be to be equipped on the External memory equipment of the electronic device 1, such as the electronic device 1 in further embodiments
Plug-in type hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card,
Flash card (Flash Card) etc..Further, the memory 11 can also both include the storage inside of the electronic device 1
Unit also includes External memory equipment.The memory 11 for store the application software for being installed on the electronic device 1 and respectively
Class data, for example, the porny identification systems 10 program code etc..The memory 11 can be also used for temporarily depositing
Store up the data that has exported or will export.
The processor 12 can be in some embodiments a central processing unit (Central Processing Unit,
CPU), microprocessor or other data processing chips, for running the program code stored in the memory 11 or processing number
According to, such as execute the porny identification systems 10 etc..
The display 13 can be light-emitting diode display, liquid crystal display, touch-control liquid crystal display in some embodiments
And OLED (Organic Light-Emitting Diode, Organic Light Emitting Diode) touches device etc..The display 13 is used
In being shown in the information handled in the electronic device 1 and for showing visual user interface, such as that identifies wait for
Identification picture belongs to picture format, the qualification result etc. of the other probability of each default picture category, the picture to be identified.The electricity
The component 11-13 of sub-device 1 is in communication with each other by system bus.
Porny identification systems 10 include at least one computer-readable instruction being stored in the memory 11, should
At least one computer-readable instruction can be executed by the processor 12, to realize each embodiment of the application.
Wherein, following steps are realized when above-mentioned porny identification systems 10 are executed by the processor 12：
Step S1 detects the picture format of the picture to be identified, and utilize training in advance after receiving picture to be identified
Picture classification model the picture to be identified is identified, export the picture to be identified and belong to one or more default figures
The probability value of piece classification；Wherein, the picture classification model is convolutional neural networks model, which is pre-
It is trained first with the default other samples pictures of picture category.
In the present embodiment, porny identification systems receive the porny identification comprising picture to be identified that user sends out
Request, for example, receive user identifies request by the porny that the terminals such as mobile phone, tablet computer, self-help terminal equipment are sent,
Such as receive the pornographic that user sends in mobile phone, tablet computer, self-help terminal equipment terminal in preassembled client
Picture identification request, or receive user and sent on the browser in the terminals such as mobile phone, tablet computer, self-help terminal equipment
The porny identification request come.
Porny identification systems utilize advance trained figure after receiving the porny identification request that user sends out
The picture to be identified received is identified in piece disaggregated model, exports the picture to be identified and belongs to each default picture classification
The probability of (porny and non-porny).The picture classification model can be first passed through in advance to being largely labeled with different default pictures
The preset quantity samples pictures of classification (porny and non-porny) be identified come constantly be trained, learn, verifying,
Optimization etc., the picture to be identified can effectively be exported by, which being trained to, belongs to each default picture classification (porny and non-
Porny) probability model.For example, depth convolutional neural networks model can be used in the picture classification model
(Convolutional Neural Network, CNN) model etc..
Meanwhile the picture format of the picture to be identified is can detect, the picture format includes black-and-white photograph and colored photograph
Piece.For example, the algorithm by judging rgb value, judges whether picture to be identified is black-and-white photograph or photochrome, and to black
White photo or photochrome are marked.Specifically, judge whether picture to be identified is black-and-white photograph or photochrome
Mode is as follows：If judging, the rgb value of pixel in photo is all 0 or 1, judges that picture to be identified is black-and-white photograph, is otherwise coloured silk
Color photo.
Step S2 judges that the picture to be identified is according to the picture format of the picture to be identified and the probability value
It is no to belong to porny.For example, since porny is generally photochrome, porny be black-and-white photograph probability compared with
Low, therefore, if judging, the picture format of picture to be identified is black and white picture, this by picture classification model output waits for
The probability that identification picture belongs to porny classification is turned down, if after the probability that the picture to be identified belongs to porny classification is turned down
Still less than predetermined threshold value (such as 50%), then judge that the picture to be identified is not belonging to porny；If the picture to be identified belongs to color
The other probability of feelings picture category is more than predetermined threshold value (such as 50%) after turning down, then judges that the picture to be identified belongs to porny.If
Judge that the picture format of picture to be identified is photochrome, the then figure to be identified directly exported the picture classification model
Piece belongs to the probability value of porny classification and predetermined threshold value (such as 60%) is compared, if being less than predetermined threshold value (such as 60%),
Then judge that the picture to be identified is not belonging to porny；If more than or equal to predetermined threshold value (such as 60%), then judge that this is to be identified
Picture belongs to porny.
Compared with prior art, the present embodiment is treated identification picture by picture classification model trained in advance and is known
Not, the probability that the picture to be identified belongs to each default picture classification such as porny and non-porny is exported；And it detects
The picture format such as black-and-white photograph or photochrome of the picture to be identified；Belong to each default figure according to the picture to be identified
The picture format of the probability of piece classification and the picture to be identified, and whether judge the picture to be identified based on preset rules
Belong to porny.Since the probability that porny is black-and-white photograph is relatively low, the present invention can recognize that picture to be identified belongs to color
The probability of feelings picture and non-porny, and carry out in conjunction with the picture format of picture to be identified such as black-and-white photograph or photochrome
Comprehensive identification, can more accurately and effectively judge whether the picture to be identified belongs to porny.Moreover, without manually into
Row detection, can carry out porny identification, effectively improve detection efficiency automatically.
In an optional embodiment, on the basis of the embodiment of above-mentioned Fig. 1, the picture classification mould trained in advance
The training step of type is as follows：
A, it is the samples pictures of the corresponding preset quantity of each default picture category setting, for each samples pictures mark pair
The default picture classification answered；Wherein, the default picture classification includes porny, sexy picture and normal picture；
B, each samples pictures are subjected to picture pretreatment to obtain the training picture for waiting for model training；
C, all trained pictures are divided into the verification collection (example of the training set and the second ratio of the first ratio (for example, 75%)
Such as, 25%)；
D, the training set training convolutional neural networks model is utilized；
E, using the accuracy rate of the convolutional neural networks model of the verification collection verification training, if accuracy rate is more than or waits
In default accuracy rate, then training terminates, using trained convolutional neural networks model as picture classification model, if alternatively, accurate
True rate is less than default accuracy rate, then increases the corresponding samples pictures quantity of each default picture classification and re-execute above-mentioned steps
B, C, D, E, until the accuracy rate of the convolutional neural networks model of training is more than or equal to default accuracy rate.
Specifically, in the present embodiment when establishing CNN picture classification models, prepare a large amount of different classes of pictures first,
Prepare 100,000 as each and be classified as pornographic, sexy, normal category samples pictures, and to the classification belonging to every pictures into rower
Note, for example, can mark as follows, 0：It is pornographic；1：It is sexy；2：Normally.Wherein, the picture of normal category may include nature,
The pictures such as art, dead volume；The picture of sexy classification may include defined being not belonging to porny but between pornographic and normal
Picture, such as draw a portrait class picture；The picture of pornographic classification is the defined picture for belonging to pornographic, is that picture is known in the present embodiment
Other emphasis.After getting out samples pictures, various pretreatments can be carried out to samples pictures, such as cut out as size unification (by picture
All resize is at 100X100 sizes) or pixel unify the training picture of equal-specification.Using pretreated trained picture and
Its classification (0 marked：It is pornographic；1：It is sexy；2：Normally), convolutional neural networks (CNN) model of training preset model structure.Example
Such as, in a kind of optional embodiment, training process is as follows：
(1) convolution kernel, deviation matrix are established；
(2) using convolution kernel, the deviation matrix established convolution algorithm is carried out with training picture；
(3) result of convolution algorithm is corrected by the activation primitive of relu；
(4) pondization is pooling operations, and pooling is that one kind contracts on the basis of retaining image major part important information
The method of small image.Pond mode includes but not limited to Mean pooling (mean value sampling), Max pooling in the present embodiment
(maximum value sampling), Overlapping (overlap sampling), L2pooling (side's sampling), Local Contrast
Normalization (normalization sampling), Stochasticpooling (sampling immediately), Def-pooling (adopt by deformation constraint
Sample) etc..
(5) above-mentioned (1)~(4) are repeated 3 times；
(6) until the cost function gradient of the model of training declines, terminate model training.
In addition, in model reaches the standard grade application process, if operation system notes abnormalities after picture such as porny, hair can be supplemented
Existing abnormal picture continues to be trained the convolutional neural networks (CNN) model, constantly to promote the convolutional neural networks
(CNN) accuracy of identification of model.
After establishing CNN picture classification models, you can carry out picture classification using CNN picture classification models.By figure to be identified
Piece is input to trained CNN picture classifications model, can export the picture to be identified by trained CNN picture classifications model
Belong to each other probability of default picture category, i.e. the picture to be identified belongs to the probability P 0 of pornographic classification, the picture category to be identified
In the probability P 1 of sexy classification, which belongs to the probability P 2 of normal category.
Meanwhile identification picture progress picture format judgement can be treated, in the present embodiment, judging that picture to be identified is black and white
On the basis of photo or photochrome, also can determine whether picture to be identified whether be the types such as dynamic picture such as GIF picture, if
Judge that the picture to be identified is dynamic picture such as GIF types, then GIF cardons are cut into frame into plurality of pictures, in case follow-up use.Tool
Body, judge picture to be identified whether be dynamic picture mode there are two types of：First, the file suffixes name for passing through picture to be identified
Judge, second is that judged by the file header of the binary format of picture to be identified.
Exporting picture to be identified using CNN picture classification models, to belong to each default picture classification (pornographic, sexy, just
Probability (P0, P1, P2) often), and judge the format (black-and-white photograph or photochrome or an action shot) of picture to be identified
Afterwards, can integrate the format of (P0, P1, P2) and picture to be identified and preset rules finally identify picture to be identified whether be
Porny.Specifically：
Since porny is generally photochrome, porny is that the probability of black-and-white photograph is relatively low, therefore, if judging
It is black and white picture to go out picture to be identified, then the picture to be identified that CNN picture classification models export is belonged to the general of pornographic classification
The value of rate P0 is turned down, for example, 10% can be adjusted downwards, and if set the P0 after turning down and be less than first threshold, judge that this waits reflecting
Determine picture and is not belonging to pornographic classification, i.e. P0*90%<50%, then do not judge that the picture to be identified is porny.If after turning down
P0 is more than first threshold, then judges that the picture to be identified belongs to pornographic classification, i.e. P0*90%>50%, then judge the figure to be identified
Piece is porny.
Meanwhile the picture to be identified that CNN picture classification models export can be belonged on sexy and normal category probability
Adjust 10%.Get picture to be identified after adjustment belong to each default picture classification (pornographic, sexy, normal) probability (P0 ',
P1 ', P2 ') after, the corresponding types of MAX (P0, P1, P2) are the type of identification, and the confidence rate of the type is MAX (P0, P1, P2),
The corresponding picture classification of maximum value (pornographic, sexy or normal) i.e. in (P0 ', P1 ', P2 ') is that this finally identified waits reflecting
Determine the classification of picture.It should be noted that on the basis of the P0 after being turned down in meeting previous step is more than first threshold, no matter
The corresponding picture classification of maximum value in (P0 ', P1 ', P2 ') is which kind of classification, the class of the picture to be identified finally identified
Porny is not remained as.
If judging, picture to be identified is dynamic picture, and the picture to be identified such as GIF cardons are cut frame into plurality of pictures,
Each frame picture to cutting frame be utilized respectively the output of CNN picture classification models belong to each default picture classification (it is pornographic, sexy,
Probability (P0, P1, P2) normally), and judge that each frame picture belongs to black-and-white photograph or photochrome, and utilize above-mentioned rule
Method carries out pornographic identification to each frame picture, according to the qualification result of each frame picture of the picture to be identified such as GIF cardons
Come whether the comprehensive judgement picture to be identified such as GIF cardons belong to porny.For example, in a kind of optional embodiment mode
In, if there is a frame picture to be accredited as pornographic classification, directly judge that the picture to be identified such as GIF cardons belong to porny.
In another optional embodiment, if there is the P0 ' of two frame pictures to be all higher than second threshold (60% or 70%), judgement should
Picture to be identified such as GIF cardons belong to porny, etc..
The result finally returned that in the present embodiment includes：Whether the picture to be identified is porny and this is to be identified
Picture belongs to the probability (P0 ', P1 ', P2 ') of each default picture classification (pornographic, sexy, normal).It is different in practical applications
Business scenario in, first threshold, second threshold can be manually set as needed, with suitable for stringent or loose porny
Identify scene.It is such as required at some in stringent scene, can be identified according to probability (P0 ', P1 ', P2 ') and belong to sexy classification
Picture.
It should be noted that when the picture format includes black and white format and non-black and white format, the default picture classification
Including porny and when non-porny, exports the picture to be identified and belong to the probability value of porny and non-porny
Probability value；Reduce the probability value that the picture that picture format is black and white format belongs to porny；By the probability after the reduction
Value is compared with the first predetermined threshold value, if it is described turn down after probability value be more than the preset first threshold value, judge described in
Picture to be identified belongs to porny；If probability value after turning down is less than or equal to the preset first threshold value, judge described in
Picture to be identified is not belonging to porny.
When the picture format includes black and white format and non-black and white format, the default picture classification include porny,
When sexy picture and normal picture, the probability that the picture to be identified belongs to the probability value of porny, sexy picture is exported
The probability value of value and normal picture；Reduce the probability value that the picture that picture format is black and white format belongs to porny；It will be described
Probability value after reduction is compared with the second predetermined threshold value, if it is described turn down after probability value be more than default second threshold
Value, then judge that the picture to be identified belongs to porny；If the probability value after turning down is less than or equal to default second threshold
Value, then judge that the picture to be identified is not belonging to porny.
Wherein, the preset first threshold value can be the same or different with the default second threshold, for example, can be arranged
The preset first threshold value is 40%, and the default second threshold is 30%.
As shown in Fig. 2, Fig. 2 is the flow diagram of one embodiment of porny identification method of the present invention, the porny
Identification method includes the following steps：
Step S10 detects the picture format of the picture to be identified, and utilize instruction in advance after receiving picture to be identified
The picture to be identified is identified in experienced picture classification model, exports the picture to be identified and belongs to one or more default
The other probability value of picture category；Wherein, the picture classification model is convolutional neural networks model, which is
Advance with what the other samples pictures of default picture category were trained.
In the present embodiment, porny identification systems receive the porny identification comprising picture to be identified that user sends out
Request, for example, receive user identifies request by the porny that the terminals such as mobile phone, tablet computer, self-help terminal equipment are sent,
Such as receive the pornographic that user sends in mobile phone, tablet computer, self-help terminal equipment terminal in preassembled client
Picture identification request, or receive user and sent on the browser in the terminals such as mobile phone, tablet computer, self-help terminal equipment
The porny identification request come.
Porny identification systems utilize advance trained figure after receiving the porny identification request that user sends out
The picture to be identified received is identified in piece disaggregated model, exports the picture to be identified and belongs to each default picture classification
The probability of (porny and non-porny).The picture classification model can be first passed through in advance to being largely labeled with different default pictures
The preset quantity samples pictures of classification (porny and non-porny) be identified come constantly be trained, learn, verifying,
Optimization etc., the picture to be identified can effectively be exported by, which being trained to, belongs to each default picture classification (porny and non-
Porny) probability model.For example, depth convolutional neural networks model can be used in the picture classification model
(Convolutional Neural Network, CNN) model etc..
Meanwhile the picture format of the picture to be identified is can detect, the picture format includes black-and-white photograph and colored photograph
Piece.For example, the algorithm by judging rgb value, judges whether picture to be identified is black-and-white photograph or photochrome, and to black
White photo or photochrome are marked.Specifically, judge whether picture to be identified is black-and-white photograph or photochrome
Mode is as follows：If judging, the rgb value of pixel in photo is all 0 or 1, judges that picture to be identified is black-and-white photograph, is otherwise coloured silk
Color photo.
Step S20 judges that the picture to be identified is according to the picture format of the picture to be identified and the probability value
It is no to belong to porny.For example, since porny is generally photochrome, porny be black-and-white photograph probability compared with
Low, therefore, if judging, the picture format of picture to be identified is black and white picture, this by picture classification model output waits for
The probability that identification picture belongs to porny classification is turned down, if after the probability that the picture to be identified belongs to porny classification is turned down
Still less than predetermined threshold value (such as 50%), then judge that the picture to be identified is not belonging to porny；If the picture to be identified belongs to color
The other probability of feelings picture category is more than predetermined threshold value (such as 50%) after turning down, then judges that the picture to be identified belongs to porny.If
Judge that the picture format of picture to be identified is photochrome, the then figure to be identified directly exported the picture classification model
Piece belongs to the probability value of porny classification and predetermined threshold value (such as 60%) is compared, if being less than predetermined threshold value (such as 60%),
Then judge that the picture to be identified is not belonging to porny；If more than or equal to predetermined threshold value (such as 60%), then judge that this is to be identified
Picture belongs to porny.
Compared with prior art, the present embodiment is treated identification picture by picture classification model trained in advance and is known
Not, the probability that the picture to be identified belongs to each default picture classification such as porny and non-porny is exported；And it detects
The picture format such as black-and-white photograph or photochrome of the picture to be identified；Belong to each default figure according to the picture to be identified
The picture format of the probability of piece classification and the picture to be identified, and whether judge the picture to be identified based on preset rules
Belong to porny.Since the probability that porny is black-and-white photograph is relatively low, the present invention can recognize that picture to be identified belongs to color
The probability of feelings picture and non-porny, and carry out in conjunction with the picture format of picture to be identified such as black-and-white photograph or photochrome
Comprehensive identification, can more accurately and effectively judge whether the picture to be identified belongs to porny.Moreover, without manually into
Row detection, can carry out porny identification, effectively improve detection efficiency automatically.
In an optional embodiment, on the basis of the above embodiments, the trained picture classification model in advance
Training step is as follows：
A, it is the samples pictures of the corresponding preset quantity of each default picture category setting, for each samples pictures mark pair
The default picture classification answered；Wherein, the default picture classification includes porny, sexy picture and normal picture；
B, each samples pictures are subjected to picture pretreatment to obtain the training picture for waiting for model training；
C, all trained pictures are divided into the verification collection (example of the training set and the second ratio of the first ratio (for example, 75%)
Such as, 25%)；
D, the training set training convolutional neural networks model is utilized；
E, using the accuracy rate of the convolutional neural networks model of the verification collection verification training, if accuracy rate is more than or waits
In default accuracy rate, then training terminates, using trained convolutional neural networks model as picture classification model, if alternatively, accurate
True rate is less than default accuracy rate, then increases the corresponding samples pictures quantity of each default picture classification and re-execute above-mentioned steps
B, C, D, E, until the accuracy rate of the convolutional neural networks model of training is more than or equal to default accuracy rate.
Specifically, in the present embodiment when establishing CNN picture classification models, prepare a large amount of different classes of pictures first,
Prepare 100,000 as each and be classified as pornographic, sexy, normal category samples pictures, and to the classification belonging to every pictures into rower
Note, for example, can mark as follows, 0：It is pornographic；1：It is sexy；2：Normally.Wherein, the picture of normal category may include nature,
The pictures such as art, dead volume；The picture of sexy classification may include defined being not belonging to porny but between pornographic and normal
Picture, such as draw a portrait class picture；The picture of pornographic classification is the defined picture for belonging to pornographic, is that picture is known in the present embodiment
Other emphasis.After getting out samples pictures, various pretreatments can be carried out to samples pictures, such as cut out as size unification (by picture
All resize is at 100X100 sizes) or pixel unify the training picture of equal-specification.Using pretreated trained picture and
Its classification (0 marked：It is pornographic；1：It is sexy；2：Normally), convolutional neural networks (CNN) model of training preset model structure.Example
Such as, in a kind of optional embodiment, training process is as follows：
(1) convolution kernel, deviation matrix are established；
(2) using convolution kernel, the deviation matrix established convolution algorithm is carried out with training picture；
(3) result of convolution algorithm is corrected by the activation primitive of relu；
(4) pondization is pooling operations, and pooling is that one kind contracts on the basis of retaining image major part important information
The method of small image.Pond mode includes but not limited to Mean pooling (mean value sampling), Max pooling in the present embodiment
(maximum value sampling), Overlapping (overlap sampling), L2pooling (side's sampling), Local Contrast
Normalization (normalization sampling), Stochasticpooling (sampling immediately), Def-pooling (adopt by deformation constraint
Sample) etc..
(5) above-mentioned (1)~(4) are repeated 3 times；
(6) until the cost function gradient of the model of training declines, terminate model training.
In addition, in model reaches the standard grade application process, if operation system notes abnormalities after picture such as porny, hair can be supplemented
Existing abnormal picture continues to be trained the convolutional neural networks (CNN) model, constantly to promote the convolutional neural networks
(CNN) accuracy of identification of model.
After establishing CNN picture classification models, you can carry out picture classification using CNN picture classification models.By figure to be identified
Piece is input to trained CNN picture classifications model, can export the picture to be identified by trained CNN picture classifications model
Belong to each other probability of default picture category, i.e. the picture to be identified belongs to the probability P 0 of pornographic classification, the picture category to be identified
In the probability P 1 of sexy classification, which belongs to the probability P 2 of normal category.
Meanwhile identification picture progress picture format judgement can be treated, in the present embodiment, judging that picture to be identified is black and white
On the basis of photo or photochrome, also can determine whether picture to be identified whether be the types such as dynamic picture such as GIF picture, if
Judge that the picture to be identified is dynamic picture such as GIF types, then GIF cardons are cut into frame into plurality of pictures, in case follow-up use.Tool
Body, judge picture to be identified whether be dynamic picture mode there are two types of：First, the file suffixes name for passing through picture to be identified
Judge, second is that judged by the file header of the binary format of picture to be identified.
Exporting picture to be identified using CNN picture classification models, to belong to each default picture classification (pornographic, sexy, just
Probability (P0, P1, P2) often), and judge the format (black-and-white photograph or photochrome or an action shot) of picture to be identified
Afterwards, can integrate the format of (P0, P1, P2) and picture to be identified and preset rules finally identify picture to be identified whether be
Porny.Specifically：
Since porny is generally photochrome, porny is that the probability of black-and-white photograph is relatively low, therefore, if judging
It is black and white picture to go out picture to be identified, then the picture to be identified that CNN picture classification models export is belonged to the general of pornographic classification
The value of rate P0 is turned down, for example, 10% can be adjusted downwards, and if set the P0 after turning down and be less than first threshold, judge that this waits reflecting
Determine picture and is not belonging to pornographic classification, i.e. P0*90%<50%, then do not judge that the picture to be identified is porny.If after turning down
P0 is more than first threshold, then judges that the picture to be identified belongs to pornographic classification, i.e. P0*90%>50%, then judge the figure to be identified
Piece is porny.
Meanwhile the picture to be identified that CNN picture classification models export can be belonged on sexy and normal category probability
Adjust 10%.Get picture to be identified after adjustment belong to each default picture classification (pornographic, sexy, normal) probability (P0 ',
P1 ', P2 ') after, the corresponding types of MAX (P0, P1, P2) are the type of identification, and the confidence rate of the type is MAX (P0, P1, P2),
The corresponding picture classification of maximum value (pornographic, sexy or normal) i.e. in (P0 ', P1 ', P2 ') is that this finally identified waits reflecting
Determine the classification of picture.It should be noted that on the basis of the P0 after being turned down in meeting previous step is more than first threshold, no matter
The corresponding picture classification of maximum value in (P0 ', P1 ', P2 ') is which kind of classification, the class of the picture to be identified finally identified
Porny is not remained as.
If judging, picture to be identified is dynamic picture, and the picture to be identified such as GIF cardons are cut frame into plurality of pictures,
Each frame picture to cutting frame be utilized respectively the output of CNN picture classification models belong to each default picture classification (it is pornographic, sexy,
Probability (P0, P1, P2) normally), and judge that each frame picture belongs to black-and-white photograph or photochrome, and utilize above-mentioned rule
Method carries out pornographic identification to each frame picture, according to the qualification result of each frame picture of the picture to be identified such as GIF cardons
Come whether the comprehensive judgement picture to be identified such as GIF cardons belong to porny.For example, in a kind of optional embodiment mode
In, if there is a frame picture to be accredited as pornographic classification, directly judge that the picture to be identified such as GIF cardons belong to porny.
In another optional embodiment, if there is the P0 ' of two frame pictures to be all higher than second threshold (60% or 70%), judgement should
Picture to be identified such as GIF cardons belong to porny, etc..
The result finally returned that in the present embodiment includes：Whether the picture to be identified is porny and this is to be identified
Picture belongs to the probability (P0 ', P1 ', P2 ') of each default picture classification (pornographic, sexy, normal).It is different in practical applications
Business scenario in, first threshold, second threshold can be manually set as needed, with suitable for stringent or loose porny
Identify scene.It is such as required at some in stringent scene, can be identified according to probability (P0 ', P1 ', P2 ') and belong to sexy classification
Picture.
It should be noted that when the picture format includes black and white format and non-black and white format, the default picture classification
Including porny and when non-porny, exports the picture to be identified and belong to the probability value of porny and non-porny
Probability value；Reduce the probability value that the picture that picture format is black and white format belongs to porny；By the probability after the reduction
Value is compared with the first predetermined threshold value, if it is described turn down after probability value be more than the preset first threshold value, judge described in
Picture to be identified belongs to porny；If probability value after turning down is less than or equal to the preset first threshold value, judge described in
Picture to be identified is not belonging to porny.
When the picture format includes black and white format and non-black and white format, the default picture classification include porny,
When sexy picture and normal picture, the probability that the picture to be identified belongs to the probability value of porny, sexy picture is exported
The probability value of value and normal picture；Reduce the probability value that the picture that picture format is black and white format belongs to porny；It will be described
Probability value after reduction is compared with the second predetermined threshold value, if it is described turn down after probability value be more than default second threshold
Value, then judge that the picture to be identified belongs to porny；If the probability value after turning down is less than or equal to default second threshold
Value, then judge that the picture to be identified is not belonging to porny.
Wherein, the preset first threshold value can be the same or different with the default second threshold, for example, can be arranged
The preset first threshold value is 40%, and the default second threshold is 30%.
In addition, the present invention also provides a kind of computer readable storage medium, the computer-readable recording medium storage has
Porny identification systems, the porny identification systems can be executed by least one processor, so that described at least one
Processor is executed such as the step of porny identification method in above-described embodiment, the step S10 of the porny identification method,
The specific implementation process such as S20, S30 are as described above, and details are not described herein.
It should be noted that herein, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that process, method, article or device including a series of elements include not only those elements, and
And further include other elements that are not explicitly listed, or further include for this process, method, article or device institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including this
There is also other identical elements in the process of element, method, article or device.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can add the mode of required general hardware platform to realize by software, naturally it is also possible to be realized by hardware, but very much
In the case of the former be more preferably embodiment.Based on this understanding, technical scheme of the present invention is substantially in other words to existing
The part that technology contributes can be expressed in the form of software products, which is stored in a storage
In medium (such as ROM/RAM, magnetic disc, CD), including some instructions are used so that a station terminal equipment (can be mobile phone, calculate
Machine, server, air conditioner or network equipment etc.) execute method described in each embodiment of the present invention.
Above by reference to the preferred embodiment of the present invention has been illustrated, not thereby limit to the interest field of the present invention.On
It is for illustration only to state serial number of the embodiment of the present invention, can not represent the quality of embodiment.It is patrolled in addition, though showing in flow charts
Sequence is collected, but in some cases, it can be with the steps shown or described are performed in an order that is different from the one herein.
Those skilled in the art do not depart from the scope of the present invention and essence, can there are many variant scheme realize the present invention,
It can be used for another embodiment for example as the feature of one embodiment and obtain another embodiment.All technologies with the present invention
All any modification, equivalent and improvement made by within design, should all be within the interest field of the present invention.
Claims (10)
1. a kind of porny identification method, which is characterized in that the porny identification method includes：
After receiving picture to be identified, the picture format of the picture to be identified is detected, and utilize picture classification trained in advance
The picture to be identified is identified in model, and it is other general that the output picture to be identified belongs to one or more default picture categories
Rate value；Wherein, the picture classification model is convolutional neural networks model, the convolutional neural networks model be advance with it is described
What the default other samples pictures of picture category were trained；
According to the picture format of the picture to be identified and the probability value, judge whether the picture to be identified belongs to pornographic figure
Piece.
2. porny identification method as described in claim 1, which is characterized in that the picture format include black and white format and
Non- black and white format, the default picture classification includes porny and non-porny；
The output picture to be identified belongs to the step of one or more default picture categories other probability value and specifically includes：It is defeated
Go out the probability value that the picture to be identified belongs to the probability value and non-porny of porny；
The picture format according to the picture to be identified and the probability value, judge whether the picture to be identified belongs to color
The step of feelings picture, specifically includes：
Reduce the probability value that the picture that picture format is black and white format belongs to porny；
Judge whether the picture of the black and white format belongs to porny according to the probability value after reduction, and according to the probability
Value judges whether the picture of non-black and white format belongs to porny.
3. porny identification method as claimed in claim 2, which is characterized in that the probability value according to after reduction judges
The step of whether picture of the black and white format belongs to porny specifically includes：
Probability value after the reduction is compared with the first predetermined threshold value, if it is described turn down after probability value be more than it is described pre-
If first threshold, then judge that the picture to be identified belongs to porny；
If the probability value after turning down is less than or equal to the preset first threshold value, judge that the picture to be identified is not belonging to pornographic
Picture.
4. porny identification method as described in claim 1, which is characterized in that the picture format include black and white format and
Non- black and white format, the default picture classification include porny, sexy picture and normal picture；
The output picture to be identified belongs to the step of one or more default picture categories other probability value and specifically includes：It is defeated
Go out probability value, the sexy probability value of picture and the probability value of normal picture that the picture to be identified belongs to porny；
The picture format according to the picture to be identified and the probability value, judge whether the picture to be identified belongs to color
The step of feelings picture, specifically includes：
Reduce the probability value that the picture that picture format is black and white format belongs to porny；
Judge whether the picture of the black and white format belongs to porny according to the probability value after reduction, and according to the probability
Value judges whether the picture of non-black and white format belongs to porny.
5. porny identification method as claimed in claim 4, which is characterized in that the probability value according to after reduction judges
The step of whether picture of the black and white format belongs to porny specifically includes：
Probability value after the reduction is compared with the second predetermined threshold value, if it is described turn down after probability value be more than it is described pre-
If second threshold, then judge that the picture to be identified belongs to porny；
If the probability value after turning down is less than or equal to the default second threshold, judge that the picture to be identified is not belonging to pornographic
Picture.
6. porny identification method as described in claim 4 or 5, which is characterized in that the training of the picture classification model
Steps are as follows：
A, porny, sexy picture and the normal picture of preset quantity are collected as samples pictures, and in each samples pictures
It is upper to mark corresponding picture classification；
B, picture pretreatment is carried out to each samples pictures；
C, pretreated samples pictures are divided into the verification collection of the training set and the second ratio of the first ratio；
D, the training set training convolutional neural networks model is utilized；
E, using the accuracy rate of the convolutional neural networks model of the verification collection verification training, if accuracy rate is more than or equal to pre-
If accuracy rate, then training terminates, and using trained convolutional neural networks model as picture classification model；If accuracy rate is less than
Default accuracy rate then collects samples pictures and repeats above-mentioned steps B, C, D, E again.
7. porny identification method as described in any one in claim 1-5, which is characterized in that when the picture to be identified is
It is further comprising the steps of when dynamic picture：
The picture to be identified is cut into frame into plurality of pictures, detection cuts the picture format of each pictures of frame, and is utilized respectively
The plurality of pictures for cutting frame is identified in trained picture classification model in advance, and each pictures that output cuts frame belong to each pre-
If the other probability value of picture category；
According to the picture format of each pictures of described section of frame and the probability value, judge whether the picture to be identified belongs to
Porny.
8. porny identification method as claimed in claim 7, which is characterized in that cut each figure of frame described in the basis
The picture format of piece and the probability value judge that the step of whether picture to be identified belongs to porny specifically includes：
According to the picture format of each pictures of described section of frame and the probability value, judge in the picture of section frame with the presence or absence of category
In the picture of porny；
If in the presence of judging that the picture to be identified belongs to porny；Otherwise, judge that the picture to be identified is not belonging to pornographic
Picture；Or
According to the picture format of each pictures of described section of frame and the probability value, judge that porny is general in the picture of section frame
Whether rate value is more than the picture of preset value more than two；
If so, the judgement picture to be identified belongs to porny；Otherwise, judge that the picture to be identified is not belonging to pornographic figure
Piece.
9. a kind of electronic device, which is characterized in that the electronic device includes memory, processor, is stored on the memory
There are the porny identification systems that can be run on the processor, the porny identification systems to be executed by the processor
The step of Shi Shixian such as claim 1-8 any one of them porny identification methods.
10. a kind of computer readable storage medium, which is characterized in that be stored with pornographic figure on the computer readable storage medium
Piece identification systems realize such as claim 1-8 any one of them colors when the porny identification systems are executed by processor
The step of feelings picture identification method.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810184583.6A CN108491866B (en) | 2018-03-06 | 2018-03-06 | Pornographic picture identification method, electronic device and readable storage medium |
| PCT/CN2018/089718 WO2019169767A1 (en) | 2018-03-06 | 2018-06-03 | Pornographic picture identification method, electronic device, and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810184583.6A CN108491866B (en) | 2018-03-06 | 2018-03-06 | Pornographic picture identification method, electronic device and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108491866A true CN108491866A (en) | 2018-09-04 |
| CN108491866B CN108491866B (en) | 2022-09-13 |
Family
ID=63341693
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810184583.6A Active CN108491866B (en) | 2018-03-06 | 2018-03-06 | Pornographic picture identification method, electronic device and readable storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN108491866B (en) |
| WO (1) | WO2019169767A1 (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109491970A (en) * | 2018-10-11 | 2019-03-19 | 平安科技（深圳）有限公司 | Imperfect picture detection method, device and storage medium towards cloud storage |
| CN109523281A (en) * | 2018-11-26 | 2019-03-26 | 北京拓世寰宇网络技术有限公司 | A kind of determining source of houses picture category method for distinguishing and device |
| CN109886865A (en) * | 2019-01-07 | 2019-06-14 | 平安科技（深圳）有限公司 | Method, apparatus, computer equipment and the storage medium of automatic shield flame |
| CN110210356A (en) * | 2019-05-24 | 2019-09-06 | 厦门美柚信息科技有限公司 | A kind of picture discrimination method, apparatus and system |
| CN110456955A (en) * | 2019-08-01 | 2019-11-15 | 腾讯科技（深圳）有限公司 | Exposure dress ornament detection method, device, system, equipment and storage medium |
| CN111199233A (en) * | 2019-12-30 | 2020-05-26 | 四川大学 | Improved deep learning pornographic image identification method |
| CN115546824A (en) * | 2022-04-18 | 2022-12-30 | 荣耀终端有限公司 | Taboo picture identification method, equipment and storage medium |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111259823A (en) * | 2020-01-19 | 2020-06-09 | 人民中科(山东)智能技术有限公司 | Pornographic image identification method based on convolutional neural network |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150010218A1 (en) * | 2012-01-11 | 2015-01-08 | 77 Elektronika Muszeripari Kft. | Image Processing Method And Apparatus |
| CN105761201A (en) * | 2016-02-02 | 2016-07-13 | 山东大学 | Method for translation of characters in picture |
| CN105989330A (en) * | 2015-02-03 | 2016-10-05 | 阿里巴巴集团控股有限公司 | Picture detection method and apparatus |
| CN106096602A (en) * | 2016-06-21 | 2016-11-09 | 苏州大学 | Chinese license plate recognition method based on convolutional neural network |
| CN106228158A (en) * | 2016-07-25 | 2016-12-14 | 北京小米移动软件有限公司 | The method and apparatus of picture detection |
| CN106874924A (en) * | 2015-12-14 | 2017-06-20 | 阿里巴巴集团控股有限公司 | A kind of recognition methods of picture style and device |
| CN106874296A (en) * | 2015-12-14 | 2017-06-20 | 阿里巴巴集团控股有限公司 | A kind of style recognition methods of commodity and device |
| CN106951848A (en) * | 2017-03-13 | 2017-07-14 | 平安科技（深圳）有限公司 | The method and system of picture recognition |
| CN107067020A (en) * | 2016-12-30 | 2017-08-18 | 腾讯科技（上海）有限公司 | Image identification method and device |
| CN107180225A (en) * | 2017-04-19 | 2017-09-19 | 华南理工大学 | A kind of recognition methods for cartoon figure's facial expression |
| CN107330453A (en) * | 2017-06-19 | 2017-11-07 | 中国传媒大学 | The Pornographic image recognizing method of key position detection is recognized and merged based on substep |
| US20180005079A1 (en) * | 2016-07-01 | 2018-01-04 | Ricoh Co., Ltd. | Active View Planning By Deep Learning |
| CN107665333A (en) * | 2017-08-28 | 2018-02-06 | 平安科技（深圳）有限公司 | A kind of indecency image identification method, terminal, equipment and computer-readable recording medium based on convolutional neural networks |
| CN107665353A (en) * | 2017-09-15 | 2018-02-06 | 平安科技（深圳）有限公司 | Model recognizing method, device, equipment and computer-readable recording medium based on convolutional neural networks |
| CN107729924A (en) * | 2017-09-25 | 2018-02-23 | 平安科技（深圳）有限公司 | Picture review probability interval generation method and picture review decision method |
-
2018
- 2018-03-06 CN CN201810184583.6A patent/CN108491866B/en active Active
- 2018-06-03 WO PCT/CN2018/089718 patent/WO2019169767A1/en active Application Filing
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150010218A1 (en) * | 2012-01-11 | 2015-01-08 | 77 Elektronika Muszeripari Kft. | Image Processing Method And Apparatus |
| CN105989330A (en) * | 2015-02-03 | 2016-10-05 | 阿里巴巴集团控股有限公司 | Picture detection method and apparatus |
| CN106874924A (en) * | 2015-12-14 | 2017-06-20 | 阿里巴巴集团控股有限公司 | A kind of recognition methods of picture style and device |
| CN106874296A (en) * | 2015-12-14 | 2017-06-20 | 阿里巴巴集团控股有限公司 | A kind of style recognition methods of commodity and device |
| CN105761201A (en) * | 2016-02-02 | 2016-07-13 | 山东大学 | Method for translation of characters in picture |
| CN106096602A (en) * | 2016-06-21 | 2016-11-09 | 苏州大学 | Chinese license plate recognition method based on convolutional neural network |
| US20180005079A1 (en) * | 2016-07-01 | 2018-01-04 | Ricoh Co., Ltd. | Active View Planning By Deep Learning |
| CN106228158A (en) * | 2016-07-25 | 2016-12-14 | 北京小米移动软件有限公司 | The method and apparatus of picture detection |
| CN107067020A (en) * | 2016-12-30 | 2017-08-18 | 腾讯科技（上海）有限公司 | Image identification method and device |
| CN106951848A (en) * | 2017-03-13 | 2017-07-14 | 平安科技（深圳）有限公司 | The method and system of picture recognition |
| CN107180225A (en) * | 2017-04-19 | 2017-09-19 | 华南理工大学 | A kind of recognition methods for cartoon figure's facial expression |
| CN107330453A (en) * | 2017-06-19 | 2017-11-07 | 中国传媒大学 | The Pornographic image recognizing method of key position detection is recognized and merged based on substep |
| CN107665333A (en) * | 2017-08-28 | 2018-02-06 | 平安科技（深圳）有限公司 | A kind of indecency image identification method, terminal, equipment and computer-readable recording medium based on convolutional neural networks |
| CN107665353A (en) * | 2017-09-15 | 2018-02-06 | 平安科技（深圳）有限公司 | Model recognizing method, device, equipment and computer-readable recording medium based on convolutional neural networks |
| CN107729924A (en) * | 2017-09-25 | 2018-02-23 | 平安科技（深圳）有限公司 | Picture review probability interval generation method and picture review decision method |
Non-Patent Citations (2)
| Title |
|---|
| 蒋艳凰等: "基于纠错编码的CSNN及其在遥感图像分类中的应用", 《计算机研究与发展》 * |
| 郑瑞东: "利用网页特征识别不良图像网页", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109491970A (en) * | 2018-10-11 | 2019-03-19 | 平安科技（深圳）有限公司 | Imperfect picture detection method, device and storage medium towards cloud storage |
| CN109491970B (en) * | 2018-10-11 | 2024-05-10 | 平安科技（深圳）有限公司 | Bad picture detection method and device for cloud storage and storage medium |
| CN109523281A (en) * | 2018-11-26 | 2019-03-26 | 北京拓世寰宇网络技术有限公司 | A kind of determining source of houses picture category method for distinguishing and device |
| CN109886865A (en) * | 2019-01-07 | 2019-06-14 | 平安科技（深圳）有限公司 | Method, apparatus, computer equipment and the storage medium of automatic shield flame |
| CN109886865B (en) * | 2019-01-07 | 2024-01-23 | 平安科技（深圳）有限公司 | Method, device, computer equipment and storage medium for automatically shielding bad information |
| CN110210356A (en) * | 2019-05-24 | 2019-09-06 | 厦门美柚信息科技有限公司 | A kind of picture discrimination method, apparatus and system |
| CN110456955A (en) * | 2019-08-01 | 2019-11-15 | 腾讯科技（深圳）有限公司 | Exposure dress ornament detection method, device, system, equipment and storage medium |
| CN111199233A (en) * | 2019-12-30 | 2020-05-26 | 四川大学 | Improved deep learning pornographic image identification method |
| CN115546824A (en) * | 2022-04-18 | 2022-12-30 | 荣耀终端有限公司 | Taboo picture identification method, equipment and storage medium |
| CN115546824B (en) * | 2022-04-18 | 2023-11-28 | 荣耀终端有限公司 | Taboo picture identification method, apparatus and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2019169767A1 (en) | 2019-09-12 |
| CN108491866B (en) | 2022-09-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108491866A (en) | Porny identification method, electronic device and readable storage medium storing program for executing | |
| CN107194398B (en) | Vehicle damages recognition methods and the system at position | |
| CN107835496B (en) | Spam short message identification method and device and server | |
| US10332245B1 (en) | Systems and methods for quality assurance of image recognition model | |
| CN109657694A (en) | Picture automatic classification method, device and computer readable storage medium | |
| CN109284733B (en) | Shopping guide negative behavior monitoring method based on yolo and multitask convolutional neural network | |
| CN112052850B (en) | License plate recognition method and device, electronic equipment and storage medium | |
| CN110135421A (en) | Licence plate recognition method, device, computer equipment and computer readable storage medium | |
| CN113705460B (en) | Method, device, equipment and storage medium for detecting open and closed eyes of face in image | |
| CN110363747A (en) | Intelligent abnormal cell judgment method, device and computer readable storage medium | |
| CN110956123B (en) | Method, device, server and storage medium for auditing rich media content | |
| CN110246506A (en) | Voice intelligent detecting method, device and computer readable storage medium | |
| CN109977750A (en) | Seal true and false method of calibration, device and computer readable storage medium | |
| CN114708539A (en) | Image type identification method and device, equipment, medium and product thereof | |
| CN109934194A (en) | Picture classification method, edge device, system and storage medium | |
| CN107784482A (en) | Recruitment methods, electronic installation and readable storage medium storing program for executing | |
| CN110232381A (en) | License Plate Segmentation method, apparatus, computer equipment and computer readable storage medium | |
| CN110503089A (en) | OCR identification model training method, device and computer equipment based on crowdsourcing technology | |
| WO2021185184A1 (en) | Content recommendation method and apparatus, electronic device, and storage medium | |
| CN113934611A (en) | Statistical method and device for access information, electronic equipment and readable storage medium | |
| CN116758373A (en) | Training method, image processing method, device and equipment for deep learning model | |
| CN113888760B (en) | Method, device, equipment and medium for monitoring violation information based on software application | |
| CN113792801B (en) | Method, device, equipment and storage medium for detecting face dazzling degree | |
| CN115700845A (en) | Face recognition model training method, face recognition device and related equipment | |
| CN108898167A (en) | It breaks one's promise the display methods and device of number |
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 |