CN110222746A - Method, apparatus, electronic equipment and the computer readable storage medium of training classifier - Google Patents
Method, apparatus, electronic equipment and the computer readable storage medium of training classifier Download PDFInfo
- Publication number
- CN110222746A CN110222746A CN201910439084.1A CN201910439084A CN110222746A CN 110222746 A CN110222746 A CN 110222746A CN 201910439084 A CN201910439084 A CN 201910439084A CN 110222746 A CN110222746 A CN 110222746A
- Authority
- CN
- China
- Prior art keywords
- chinese character
- image
- card
- classifier
- convolutional neural
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2431—Multiple classes
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The present disclosure discloses a kind of methods of training convolutional neural networks classifier, it is characterized in that, it include: to obtain ID Card Image set, ID Card Image in the ID Card Image set is corresponding with Chinese character label information, and the Chinese character label information is used to indicate Chinese character included by the predeterminable area of ID Card Image corresponding with the Chinese character label information;Determine that the output project of convolutional neural networks classifier, the output project of the convolutional neural networks classifier are corresponding with the Chinese character label information;According to the ID Card Image set training convolutional neural networks classifier.Method, apparatus, electronic equipment and the computer readable storage medium for the training convolutional neural networks classifier that the embodiment of the present disclosure provides, it can be according to ID Card Image set training convolutional neural networks classifier, the Chinese character such as rare Chinese character in predeterminable area to identify ID Card Image by the convolutional neural networks classifier, to be applied to various ID card verification occasions.
Description
Technical field
This disclosure relates to field of information processing more particularly to a kind of method, apparatus, electronic equipment and the meter of trained classifier
Calculation machine readable storage medium storing program for executing.
Background technique
With the progress of network technology, internet has become the important medium for serving various businesses, such as by mutual
When the important services such as networking processing banking, need to obtain the ID Card Image of user, to pass through ID Card Image
The identity information of user is extracted to be used for ID card verification, to carry out important service.
In above-mentioned verification process, the ID Card Image that is uploaded often through identification device and/or recognizer from user
Then middle extraction ID Card Image region identifies the text in predeterminable area, to extract user in ID Card Image region
The information such as identity information, such as name, address.But the quantity of the Chinese character in being used be it is very huge, according to statistics,
In the Chinese character that identity card is related to, common Chinese character substantially has more than 4000, and uncommon Chinese character substantially has more than 5000, existing
Identification device and/or recognizer most of rare Chinese characters in ID Card Image can not be accomplished to accurately identify, thus
Also just the important services such as banking can not be handled by internet, brings obstruction for Internet application.
Summary of the invention
Method, apparatus, electronic equipment and the computer readable storage medium for the training classifier that the embodiment of the present disclosure provides,
It can be according to ID Card Image set training classifier, to identify ID Card Image by the convolutional neural networks classifier
Chinese character such as rare Chinese character in predeterminable area, to be applied to various ID card verification occasions.
In a first aspect, the embodiment of the present disclosure provides a kind of method of trained classifier characterized by comprising obtain body
Part demonstrate,proves image collection, and the ID Card Image in the ID Card Image set is corresponding with Chinese character label information, the Chinese character label
Information is used to indicate Chinese character included by the predeterminable area of ID Card Image corresponding with the Chinese character label information;Determine convolution
The output project of neural network classifier, the output project of the convolutional neural networks classifier and the Chinese character label information pair
It answers;According to the ID Card Image set training convolutional neural networks classifier.
Further, the ID Card Image set includes that the first subclass and second subset are closed, first subclass
In the corresponding Chinese character label information of ID Card Image indicated by the Chinese character include Chinese characters in common use, the second subset
The Chinese character indicated by the corresponding Chinese character label information of ID Card Image in conjunction includes rare Chinese character.
Further, according to the ID Card Image set training convolutional neural networks classifier, comprising: from described
The ID Card Image of the first quantity is determined in first subclass;The identity card figure of the second quantity is determined from second subset conjunction
Picture;The convolutional neural networks are updated according to the ID Card Image of first quantity and the ID Card Image of second quantity
The parameter of classifier.
Further, first quantity is 96, and second quantity is 32.
Further, the acquisition ID Card Image set, including obtain the second subset and close;Obtain second son
Set, comprising: obtain identity card template image;Obtain rare Chinese character image;According to the rare Chinese character image and the identity
Card template image synthesizes the ID Card Image in the second subset conjunction.
Further, the acquisition rare Chinese character image, comprising: the rare Chinese character image is obtained based on No. 5 words of the Song typeface,
The color of rare Chinese character in the rare Chinese character image is black, and the background color of the rare Chinese character image is white;Described
The ID Card Image in the second subset conjunction is synthesized according to the rare Chinese character image and the identity card template image, comprising:
The rare Chinese character image is covered to the predeterminable area of the identity card template image.
Further, the predeterminable area of the identity card template image includes identity card shading.
Further, the rare Chinese character image is covered to the predeterminable area of the identity card template image, packet
It includes: the rare Chinese character of the black in the rare Chinese character image is covered to the preset areas of the identity card template image
Domain.
Second aspect, the embodiment of the present disclosure provide a kind of device of trained classifier characterized by comprising image set
It closes and obtains module, ID Card Image and classification information for obtaining ID Card Image set, in the ID Card Image set
Corresponding, the classification information indicates the classification of ID Card Image corresponding with the classification information;Determining module, for determining volume
The output project of product neural network classifier, the output project are corresponding with the classification information;Training module, for according to institute
State ID Card Image set and convolutional neural networks (CNN) the training convolutional neural networks classifier.
Further, the ID Card Image set includes that the first subclass and second subset are closed, first subclass
In the corresponding Chinese character label information of ID Card Image indicated by the Chinese character include Chinese characters in common use, the second subset
The Chinese character indicated by the corresponding Chinese character label information of ID Card Image in conjunction includes rare Chinese character.
Further, the training module is also used to: the identity card figure of the first quantity is determined from first subclass
Picture;The ID Card Image of the second quantity is determined from second subset conjunction;According to the ID Card Image of first quantity and
The ID Card Image of second quantity updates the parameter of the convolutional neural networks classifier.
Further, first quantity is 96, and second quantity is 32.
Further, the acquisition module is also used to: being obtained the second subset and is closed;It obtains the second subset to close, packet
It includes: obtaining identity card template image;Obtain rare Chinese character image;According to the rare Chinese character image and the identity card Prototype drawing
As synthesizing the ID Card Image in the second subset conjunction.
Further, the acquisition module is also used to: obtaining the rare Chinese character image, the life based on No. 5 words of the Song typeface
The color of rare Chinese character in out-of-the-way Chinese character image is black, and the background color of the rare Chinese character image is white;It is described according to
Rare Chinese character image and the identity card template image synthesize the ID Card Image in the second subset conjunction, comprising: will be described
Rare Chinese character image is covered to the predeterminable area of the identity card template image.
Further, the predeterminable area of the identity card template image includes identity card shading.
Further, the rare Chinese character image is covered to the predeterminable area of the identity card template image, packet
It includes: the rare Chinese character of the black in the rare Chinese character image is covered to the preset areas of the identity card template image
Domain.
The third aspect, the embodiment of the present disclosure provide a kind of electronic equipment, comprising: memory, it is computer-readable for storing
Instruction;And the one or more processors coupled with the memory, for running the computer-readable instruction, so that institute
State the method that any trained classifier in aforementioned first aspect is realized when processor operation.
Fourth aspect, the embodiment of the present disclosure provide a kind of non-transient computer readable storage medium, which is characterized in that described
Non-transient computer readable storage medium stores computer instruction, when the computer instruction is computer-executed, so that institute
The method for stating any trained classifier that computer executes in aforementioned first aspect.
The present disclosure discloses method, apparatus, electronic equipment and the computer readable storage mediums of a kind of trained classifier.Its
Described in training classifier method characterized by comprising obtain ID Card Image set, the ID Card Image set
In ID Card Image it is corresponding with Chinese character label information, the Chinese character label information is used to indicate and the Chinese character label information pair
Chinese character included by the predeterminable area for the ID Card Image answered;Determine the output project of convolutional neural networks classifier, the volume
The output project of product neural network classifier is corresponding with the Chinese character label information;According to ID Card Image set training institute
State convolutional neural networks classifier.Method, apparatus, electronic equipment and the computer for the training classifier that the embodiment of the present disclosure provides
Readable storage medium storing program for executing, can be according to ID Card Image set training convolutional neural networks classifier, to pass through the convolutional Neural
Network classifier identifies the Chinese character such as rare Chinese character in the predeterminable area of ID Card Image, to be applied to various identity results
Demonstrate,prove occasion.
Above description is only the general introduction of disclosed technique scheme, in order to better understand the technological means of the disclosure, and
It can be implemented in accordance with the contents of the specification, and to allow the above and other objects, features and advantages of the disclosure can be brighter
Show understandable, it is special below to lift preferred embodiment, and cooperate attached drawing, detailed description are as follows.
Detailed description of the invention
In order to illustrate more clearly of the embodiment of the present disclosure or technical solution in the prior art, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is this public affairs
The some embodiments opened for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is the flow chart of the embodiment of the method for the training classifier that the embodiment of the present disclosure provides;
Fig. 2 is a kind of convolutional neural networks model schematic that the disclosure provides;
Fig. 3 is the structural schematic diagram of the Installation practice for the training classifier that the embodiment of the present disclosure provides;
Fig. 4 is the structural schematic diagram of the electronic equipment provided according to the embodiment of the present disclosure.
Specific embodiment
Illustrate embodiment of the present disclosure below by way of specific specific example, those skilled in the art can be by this specification
Disclosed content understands other advantages and effect of the disclosure easily.Obviously, described embodiment is only the disclosure
A part of the embodiment, instead of all the embodiments.The disclosure can also be subject to reality by way of a different and different embodiment
It applies or applies, the various details in this specification can also be based on different viewpoints and application, in the spirit without departing from the disclosure
Lower carry out various modifications or alterations.It should be noted that in the absence of conflict, the feature in following embodiment and embodiment can
To be combined with each other.Based on the embodiment in the disclosure, those of ordinary skill in the art are without creative efforts
Every other embodiment obtained belongs to the range of disclosure protection.
It should be noted that the various aspects of embodiment within the scope of the appended claims are described below.Ying Xian
And be clear to, aspect described herein can be embodied in extensive diversified forms, and any specific structure described herein
And/or function is only illustrative.Based on the disclosure, it will be understood by one of ordinary skill in the art that one described herein
Aspect can be independently implemented with any other aspect, and can combine the two or both in these aspects or more in various ways.
For example, carry out facilities and equipments in terms of any number set forth herein can be used and/or practice method.In addition, can make
With other than one or more of aspect set forth herein other structures and/or it is functional implement this equipment and/or
Practice the method.
It should also be noted that, diagram provided in following embodiment only illustrates the basic structure of the disclosure in a schematic way
Think, component count, shape and the size when only display is with component related in the disclosure rather than according to actual implementation in diagram are drawn
System, when actual implementation kenel, quantity and the ratio of each component can arbitrarily change for one kind, and its assembly layout kenel can also
It can be increasingly complex.
In addition, in the following description, specific details are provided for a thorough understanding of the examples.However, fields
The skilled person will understand that the aspect can be practiced without these specific details.
The method of the training classifier provided in this embodiment can be executed by the device of a trained classifier, the dress
It sets and can be implemented as software, can be implemented as hardware, be also implemented as the combination of software and hardware, such as the training classification
The device of device includes computer equipment, to execute the training classifier provided in this embodiment by the computer equipment
Method, as understood by those skilled in the art, computer equipment can be desk-top or portable computer device, can also be shifting
Dynamic terminal device etc..It, in the disclosure can be in addition, the classifier that is related to of the embodiment of the present disclosure is convolutional neural networks classifier
Abbreviation classifier.
Fig. 1 is the flow chart of the embodiment of the method one for the training classifier that the embodiment of the present disclosure provides, as shown in Figure 1, this
The method of the training classifier of open embodiment includes the following steps:
Step S101 obtains ID Card Image set, ID Card Image and Chinese character mark in the ID Card Image set
It is corresponding to sign information, the Chinese character label information is used to indicate the preset areas of ID Card Image corresponding with the Chinese character label information
Chinese character included by domain;
It needs to gather training classifier by training, therefore obtains ID Card Image set in step s101 as training
Set is to train classifier.
Since the embodiment of the present disclosure wishes to identify the Chinese character in ID Card Image by training convolutional neural networks classifier,
Such as name and address etc. in identification ID Card Image, so that the identity information of user is obtained, for use in various important industry
Business, therefore for the ID Card Image set as training set, the ID Card Image in the ID Card Image set need with
Chinese character label information is corresponding, and the Chinese character label information is used to indicate Chinese character included in the predeterminable area of ID Card Image,
To during training convolutional neural networks classifier, can by the recognition result of convolutional neural networks classifier with it is described
Chinese character label information is compared, to update the parameter of convolutional neural networks classifier, realizes training goal.
Optionally, the predeterminable area includes the name region of ID Card Image and/or the address region of ID Card Image.
There is the size of standard and the project information of standard due to identity card, and the region of projects information is also fixation, therefore
The position of the predeterminable area is also fixed.As an example, the predeterminable area includes in the name region after surname
The region of first Chinese character of name, it should be noted that name may include the feelings of two words, three words or four words
Condition, different according to the number of words of name, the region of first Chinese character of the name in the name region after surname may be
Change, but as those skilled in the art institute is specific, for the different situations of two words, three words or four words,
The region of first Chinese character of the name after surname is also fixed and is available.
Optionally, the ID Card Image in the ID Card Image set includes treated ID Card Image.Such as
The ID Card Image that user uploads may be voluntarily can have some deformation, additionally, due to bat by shootings such as mobile terminals
Environment difference is taken the photograph, light and readability etc. can may also have a difference, and the training as convolutional neural networks classifier
Set, it is desirable to ID Card Image the size more standard, clarity in acquired ID Card Image set in step s101
More high, this is beneficial to the accuracy for promoting trained convolutional neural networks classifier, therefore acquired in step S101
ID Card Image set in ID Card Image may include in treated ID Card Image, such as the step S101
Obtaining the ID Card Image set includes the image uploaded for user, is carried out by the image procossings such as stretching, noise reduction mode
Processing, so that the ID Card Image in the ID Card Image set obtained is more applicable for training convolutional neural networks classifier.
In the embodiment of the present disclosure, the Chinese character label information is used to indicate identity card corresponding with the Chinese character label information
Chinese character included by the predeterminable area of image, such as it is used to indicate the pre- of ID Card Image corresponding with the Chinese character label information
If which Chinese character the Chinese character that region is described or shown is specifically.As an example, existing as described in disclosure background technique
The Chinese character (Chinese character involved in the project that name and address including identity card etc. pass through Chinese character description) that some identity cards are related to
In, common Chinese character substantially has more than 4000, and uncommon Chinese character substantially has more than 5000, then for total more than 9000
Each of common Chinese characters in common use and uncommon rare Chinese character Chinese character configure respective Chinese character label information, total 9000
Multiple Chinese character label informations, thus according to ID Card Image and community and the Chinese character describing or show, by the identity
Demonstrate,prove one that image corresponds in more than above-mentioned 9000 a label informations;It is obtained as another example, such as in step s101
The ID Card Image set include M ID Card Image, each of M ID Card Image ID Card Image with it is N number of
A correspondence (such as M is far longer than N, such as M=500*N) in Chinese character label information, it means that for M identity card figure
An ID Card Image as in, corresponding to Chinese character label information indicated by Chinese character and one ID Card Image
Chinese character in predeterminable area is consistent.
Those skilled in the art can define, the identity card figure in the ID Card Image set in the embodiment of the present disclosure
As corresponding with the Chinese character label information, it is meant that ID Card Image and the Chinese character label in the ID Card Image set
There are corresponding relationships between information, can pass through in computer program realization as the example for not limiting the embodiment of the present disclosure
The data structures such as array, table are corresponding with the Chinese character label information by ID Card Image, for use in training convolutional neural networks
Classifier.Also, it will be understood by those skilled in the art that should make to describe in the predeterminable area of ID Card Image or show as far as possible
The corresponding Chinese character label information of the Chinese character shown matches, thus could be in the process of training convolutional neural networks classifier
In, so that neural network convolutional neural networks classifier correctly learns to have corresponded to the ID Card Image of different Chinese character label information
In describe in the predeterminable area or the characteristics of image of the Chinese character of display, so that it is higher to guarantee that convolutional neural networks classifier has
Accuracy.
Step S102, determines the output project of convolutional neural networks classifier, the convolutional neural networks classifier it is defeated
Project is corresponding with the Chinese character label information out;
As understood by one of ordinary skill in the art, as the ID Card Image set of training set, body therein
The quantity of Chinese character in the predeterminable area of the ID Card Image of Chinese character label information instruction corresponding to part card image, and is instructed according to this
Practicing the output project for gathering the classification results of trained convolutional neural networks classifier has corresponding relationship, in usual situation
Under, the output item purpose quantity of the classification results of convolutional neural networks classifier is equal to the identity card acquired in step s101
The quantity of Chinese character indicated by the corresponding Chinese character label information of ID Card Image in image collection, referring to aforementioned exemplary, in step
The ID Card Image set obtained in rapid S101 includes M ID Card Image, each of M ID Card Image identity
Card image is corresponding with one in N number of Chinese character label information, and N number of Chinese character label information has indicated respectively N number of Chinese character, then according to this
The classification results for the convolutional neural networks classifier that image collection is trained include N number of output project.
The output project of convolutional neural networks classifier in step S102 is for indicating convolutional neural networks classifier
Classification results, as understood by those skilled in the art, not according to calculation used by convolutional neural networks classifier
Together, the output project of convolutional neural networks classifier can be by diversified forms presentation class as a result, optional real as one
Example is applied, the value of some output project output is A (such as A=1) in each output project of convolutional neural networks classifier, other
The value of output project output is B (such as B=0), then the output project for meaning that the value of output is A indicates convolutional neural networks
Classification of the classifier to input picture, as an example, the output project of convolutional neural networks classifier includes N number of output project,
For being input to an ID Card Image (or input picture) of convolutional neural networks classifier, in N number of output project
An output item purpose output valve be 1, described N number of other output valves of output item purpose are 0, and one output project is corresponding
Chinese character label information indicate the Chinese character in the predeterminable area of corresponding ID Card Image (or input picture) be " opening ",
So according to the output item purpose output valve of the convolutional neural networks classifier it is found that the convolutional neural networks classifier is by the body
Chinese character in the predeterminable area of part card image (either input picture) classifies or is identified as " opening ";It is optional real as another
Apply example, the output item purpose output valve of the convolutional neural networks classifier and be 1, as an example, such as training is described
The value that each output project of convolutional neural networks classifier is exported is more than or equal to 0 and less than or equal to 1, described
The value that each output project of convolutional neural networks classifier is exported and be 1, i.e. each of convolutional neural networks classifier is defeated
The value that project is exported out represents the convolutional neural networks classifier to the class probability of input picture, and probability value is maximum defeated
Project indicates classification of the convolutional neural networks classifier to input picture out.
Step S103, according to the ID Card Image set training convolutional neural networks classifier.
The ID Card Image set of the training convolutional neural networks classifier is determined in step s101, in step
The output project of the convolutional neural networks classifier has been determined in S102, therefore in step s 103 can be according to the identity card
The image collection training convolutional neural networks classifier, the output project of the convolutional neural networks classifier are included in described
The output project of identified convolutional neural networks classifier in step S102.
Optionally, the convolutional neural networks include but is not limited to MobileNetV2, AlexNet, googleNet,
VGGNet, DenseNet etc..
It will be appreciated by those skilled in the art that different convolutional neural networks have different frameworks, this is embodied in possibility
Layer including different layer and different number.As shown in Fig. 2, the frame of typical convolutional neural networks include convolutional layer, it is non-
Linear layer, pond layer and it is fully connected layer.
Convolutional layer is mainly used for extracting characteristics of image from input picture, can be by one or more filters (also referred to as
Characteristic detector) according to preset step-length characteristics of image is extracted from input picture.As understood by those skilled in the art, scheme
As being made of pixel, each of image pixel, such as input picture packet can be characterized by color parameter and location parameter
48*48 pixel is included, is exported according to step-length by the 1 available convolutional layer of extractor characteristics of image by the filter of 5*5
The image characteristic matrix of 44*44.
It can connect non-linear layer or pond layer after convolutional layer, wherein non-linear layer is used for the image exported to convolutional layer
Feature carries out Further Feature Extraction, pond layer can using be averaged by the way of pond or the mode in maximum pond to convolutional layer or
The output result of non-linear layer is handled, and can reduce the dimension of characteristics of image, reduces operation times.
The last of convolutional neural networks is to be fully connected layer, and the last layer for being fully connected layer is output layer, can also be claimed
For the output layer of convolutional neural networks classifier, it is fully connected the characteristics of image of the layer before layer receives, and to described image spy
Sign is handled layer by layer, and finally, treated characteristics of image is input to output layer, by activation primitive to this in output layer
Characteristics of image is calculated, and calculated result is mapped to multiple output projects included by output layer, multiple output project
It can be used as the output project of convolutional neural networks classifier.
During being based on convolutional neural networks training convolutional neural networks classifier, for the figure in training set
Picture can be input to convolutional neural networks, calculate and handle layer by layer according to the framework of convolutional neural networks, be finally fully connected layer
Output layer output category result, be then compared according to the classification results with the label information of image to construct loss letter
Number, by gradient decline scheduling algorithm according to the loss function update training process involved in weight and bias etc. parameters it
Afterwards, classification results are recalculated further according to updated parameter, such iteration is completed after obtaining optimal classification results to volume
The training of product neural network classifier can be divided hence for the image of input by the convolutional neural networks classifier
Class and/or identification.
Therefore in step s 103, in the manner described above, the ID Card Image in ID Card Image set is input to volume
Product neural network by each layer of feature extraction and is calculated with training convolutional neural networks classifier, to pass through the training
Chinese character in the predeterminable area of the convolutional neural networks classifier identification input picture of completion, such as according to example above-mentioned,
The ID Card Image set acquired in step s101 includes M ID Card Image, each of M ID Card Image body
Part card image is corresponding with one in N number of Chinese character label information, correspondingly, the output project of the convolutional neural networks classifier
Including N number of output project, during training, for an ID Card Image in M ID Card Image, according to convolution mind
Through network classifier to an ID Card Image about N number of output item purpose output valve, an identity card figure can be identified
Which Chinese character Chinese character in the predeterminable area of picture is specially, then by recognition result Chinese character corresponding with an ID Card Image
Label information compares, with for constructing loss function and updating the parameter etc. of convolutional neural networks classifier, specific process can
With referring to Fig. 2 and accordingly for the description of training process, details are not described herein again.
Optionally, the ID Card Image set includes that the first subclass and second subset are closed, in first subclass
The corresponding Chinese character label information of ID Card Image indicated by the Chinese character include Chinese characters in common use, the second subset is closed
In the corresponding Chinese character label information of ID Card Image indicated by the Chinese character include rare Chinese character.
As described in disclosure background technique, existing identification device and/or recognizer are for big in ID Card Image
Most rare Chinese characters can not be accomplished to accurately identify, therefore the training convolutional neural networks of optional embodiment provided by the disclosure
The method of classifier, it is desirable to which the convolutional neural networks classifier trained can be realized Chinese character including Chinese characters in common use and the uncommon Chinese
The identification of word, therefore in above-mentioned optional embodiment, it include the first subclass in the ID Card Image set, wherein described
The Chinese character indicated by the corresponding Chinese character label information of ID Card Image in first subclass includes Chinese characters in common use, this
Mean to describe in the predeterminable area for the ID Card Image that this first son is concentrated or the Chinese character of display includes Chinese characters in common use, institute
Stating the Chinese character indicated by the corresponding Chinese character label information of ID Card Image in second subset conjunction includes abnormal Chinese character,
This means that the Chinese character described or shown in the predeterminable area for the ID Card Image that this second son is concentrated includes rare Chinese character,
That is, not only to include in the ID Card Image set that the training as training convolutional neural networks classifier is gathered
Chinese characters in common use will include also rare Chinese character, so that the convolutional neural networks classifier trained can be to the identity card of input
Chinese characters in common use and rare Chinese character in image are identified.
It is worth noting that being flexible, different period, difference for the criteria for classifying of Chinese characters in common use and rare Chinese character
Application scenarios etc. may have different divisions, it will be understood by those skilled in the art that commonly using the Chinese involved in the disclosure
Word and rare Chinese character can be divided according to any criteria for classifying, and the disclosure does not limit this, as an example, for example can be with
The Chinese character label information of ID Card Image in the ID Card Image set acquired in step s101 is counted, for
The multiple Chinese characters indicated by Chinese character label information, the quantity of the Chinese character label information of the multiple Chinese character is arranged as indicated
Sequence, using the Chinese character of most the first ratios (such as 40%) indicated by Chinese character label information as Chinese characters in common use, others are made
For rare Chinese character, or the Chinese character of first threshold reached as Chinese characters in common use using the number indicated by the Chinese character label information,
Others are used as rare Chinese character.
In an alternative embodiment, step S103: according to the ID Card Image set training convolutional Neural
Network classifier, comprising: the ID Card Image of the first quantity is determined from first subclass;From second subset conjunction
Determine the ID Card Image of the second quantity;According to the identity card figure of the ID Card Image of first quantity and second quantity
Parameter as updating the convolutional neural networks classifier.Optionally, first quantity is 96, and second quantity is 32.
As previously mentioned, for the image in training set, can be inputted during training convolutional neural networks classifier
To convolutional neural networks, the classification results then output it are compared with the Chinese character label information of image to construct loss letter
Number to decline scheduling algorithm according to the parameter of loss function update convolutional neural networks classifier by gradient, such as is trained
The parameters such as the weight of convolutional neural networks classifier involved in process and biasing.But it is specific to the embodiment of the present disclosure, it is desirable to
By the Chinese character in the predeterminable area of trained convolutional network convolutional neural networks classifier identification input ID Card Image, and just
Chinese character quantity in use is very huge, therefore this disclosure relates to convolutional network convolutional neural networks classifier output item
Purpose quantity is also very huge, such as aforementioned exemplary, the output item purpose quantity of the convolutional neural networks classifier may with to know
The quantity of other Chinese character as many, therefore in the training process, is difficult to accomplish according to ID Card Image acquired in step s101
The classification results of whole ID Card Images in set update convolution mind to construct loss function and decline scheduling algorithm according to gradient
Parameter through network classifier, therefore in the related art of training convolutional neural networks classifier, it can pass through
Batchsize parameter limits the quantity of the ID Card Image of the parameter for updating the convolutional neural networks classifier, also
It is to say during the parameter of above-mentioned update convolutional neural networks classifier, can be chosen from the ID Card Image set
Batchsize ID Card Image constructs loss function according to the classification results of batchsize ID Card Image of selection
And then the parameter that scheduling algorithm updates convolutional neural networks classifier is declined by gradient.It is optional real as provided by the disclosure
The method for applying the training convolutional neural networks classifier of example, it is desirable to which the convolutional neural networks classifier trained can be realized the Chinese
Word, the identification including Chinese characters in common use and rare Chinese character, therefore during training convolutional neural networks classifier, the disclosure
Optional embodiment creatively proposes, for updating batchsize identity card figure of convolutional neural networks classifier parameters
As including from the ID Card Image of the first quantity of determination in first subclass and the determination from second subset conjunction
The ID Card Image of second quantity, that is to say, that for updating batchsize identity of convolutional neural networks classifier parameters
It demonstrate,proves in image, not only includes describing or showing the ID Card Image of Chinese characters in common use in predeterminable area, further include predeterminable area
Middle description shows the ID Card Image of rare Chinese character, so that in the mistake for updating convolutional neural networks classifier parameters
The characteristics of image of Chinese characters in common use is not only considered in journey, and considers the characteristics of image of rare Chinese character, the convolution mind thus trained
Preferable identification or classifying quality are provided with for Chinese characters in common use and rare Chinese character through network classifier.
In another optional embodiment, step S101: ID Card Image set is obtained, comprising: obtain described second
Subclass;Wherein,
It obtains the second subset to close, comprising: obtain identity card template image;Obtain rare Chinese character image;According to described
Rare Chinese character image and the identity card template image synthesize the ID Card Image in the second subset conjunction.
In order to obtain good classifying quality, for each output project of convolutional neural networks classifier, or for
Each Chinese character label information, requires corresponding a large amount of ID Card Image with for training, right as specific example
In each Chinese character for wishing that convolutional neural networks classifier can identify, it is intended to guarantee that there are corresponding in training set
A large amount of ID Card Images, by experiment, for each Chinese character label information, in training set there are it is corresponding about
In the case where 500 ID Card Images, good classification or recognition effect can be obtained, but rare Chinese character is originally just seldom
See, therefore ID Card Image of the acquisition including rare Chinese character is even more extremely difficult, obtains the body including rare Chinese character to reduce
The cost of part card image enriches the content of the training set for training convolutional neural networks classifier, and the disclosure proposes above-mentioned
Optional embodiment, synthesized by the image of rare Chinese character with identity card template image the second subset close in identity card figure
Picture.It is worth noting that two images, which are synthesized an image, can be based on existing image processing techniques, the embodiment of the present disclosure
Which kind of do not limited for appropriately sized rare Chinese character image is synthesized on identity card template image using image composing technique
It is fixed.
Optionally, the acquisition rare Chinese character image, comprising: obtain the rare Chinese character image, institute based on No. 5 words of the Song typeface
The color for stating the rare Chinese character in rare Chinese character image is black, and the background color of the rare Chinese character image is white;The basis
The rare Chinese character image and the identity card template image synthesize the ID Card Image in the second subset conjunction, comprising: will
The rare Chinese character image is covered to the predeterminable area of the identity card template image.Identity card is used as with authoritative
Certificate, the project information of size and standard with standard, the name region relative to the actual size of identity card, on identity card
It is similar with No. 5 words of the Song typeface with the Chinese character in the region of address, and be black, therefore the life can be obtained based on No. 5 words of the Song typeface
The color of out-of-the-way Chinese character image, the rare Chinese character in the rare Chinese character image is black, it will be understood by those skilled in the art that Song
No. 5 words of body have corresponding size, therefore can generate the rare Chinese character image according to the boundary rectangle of No. 5 words of the Song typeface, i.e.,
The size of the rare Chinese character image is identical as the size of boundary rectangle of No. 5 words of the Song typeface, can also be according to tff character library text
Part directly exports the image of No. 5 words of the Song typeface, can generate the location information of output word when exporting No. 5 words of the Song typeface by tff character library,
To obtain the rare Chinese character image according to the location information, how the disclosure is for obtain the side of the rare Chinese character image
Formula is without limitation.
Optionally, the predeterminable area of the identity card template image includes identity card shading.Due to true identity
Card has shading, and can be reacted shading on the image based on ID Card Image captured by true identity card, such as black
Text under there are shadings, therefore the identity card template image in the embodiment of the present disclosure for synthesizing ID Card Image may include
Identity card shading, to obtain the ID Card Image more really synthesized.
In above-mentioned optional embodiment, since the background color of the rare Chinese character image is white, and identity card itself has
Have a shading, therefore when the predeterminable area in the identity card template image has shading, by the rare Chinese character image cover to
White-based color will be presented in the predeterminable area of the identity card template image, the Chinese character in the predeterminable area, this and body
The characteristics of image of part card has differences, and the ID Card Image training convolutional neural networks classifier based on this synthesis may
Training result is impacted.It, will be described in another optional embodiment of the disclosure in order to solve this technical problem
Rare Chinese character image is covered to the predeterminable area of the identity card template image, comprising: will be in the rare Chinese character image
The rare Chinese character of black cover to the predeterminable area of the identity card template image.In this embodiment, due to uncommon
Rare Chinese character in Chinese character image is black, and its background color is white, therefore can be based on the face of the rare Chinese character image
Color characteristic extracts the location information of black picture element, and covers to the predeterminable area of the identity card template image,
The predeterminable area of ID Card Image to synthesize is not in the Chinese character of white-based color, but occurs having identity card shading
Black Chinese character.
It is worth noting that due to the more difficult acquisition of ID Card Image that predeterminable area includes rare Chinese character, using this
The mode that ID Card Image is synthesized in open embodiment, classifies in order to obtain at lower cost for training convolutional neural networks
The training set of device, but the identity card that predeterminable area includes Chinese characters in common use cannot be obtained by above embodiment by not representing
Image, that is to say, that obtain ID Card Image set described in step S101, including obtain first subclass, obtain institute
State the first subclass, comprising: obtain identity card template image;Obtain Chinese characters in common use image;According to the Chinese characters in common use image and
The identity card template image synthesizes the ID Card Image in first subclass.Specific synthesis mode is referred to
About the identical or corresponding description for synthesizing the ID Card Image in the second subset conjunction, details are not described herein again.
Fig. 3 show the knot of 300 embodiment of device of the training convolutional neural networks classifier of embodiment of the present disclosure offer
Structure schematic diagram, as shown in figure 3, the device 300 of the training convolutional neural networks classifier includes that image collection obtains module
301, determining module 302, training module 303.
Wherein, the acquisition module 301, the body for obtaining ID Card Image set, in the ID Card Image set
Part card image is corresponding with Chinese character label information, and the Chinese character label information is used to indicate body corresponding with the Chinese character label information
Chinese character included by the predeterminable area of part card image;The determining module 302, for determining the defeated of convolutional neural networks classifier
Project out, the output project of the convolutional neural networks classifier are corresponding with the Chinese character label information;The training module
303, for according to the ID Card Image set training convolutional neural networks classifier.
The method that Fig. 3 shown device can execute embodiment illustrated in fig. 1, the part that the present embodiment is not described in detail can join
Examine the related description to embodiment illustrated in fig. 1.In implementation procedure and the technical effect embodiment shown in Figure 1 of the technical solution
Description, details are not described herein.
Below with reference to Fig. 4, it illustrates the structural representations for the electronic equipment 500 for being suitable for being used to realize the embodiment of the present disclosure
Figure.Electronic equipment in the embodiment of the present disclosure can include but is not limited to such as mobile phone, laptop, digital broadcasting and connect
Receive device, PDA (personal digital assistant), PAD (tablet computer), PMP (portable media player), car-mounted terminal (such as vehicle
Carry navigation terminal) etc. mobile terminal and such as number TV, desktop computer etc. fixed terminal.Electricity shown in Fig. 4
Sub- equipment is only an example, should not function to the embodiment of the present disclosure and use scope bring any restrictions.
As shown in figure 4, electronic equipment 400 may include processing unit (such as central processing unit, graphics processor etc.)
401, random access can be loaded into according to the program being stored in read-only memory (ROM) 402 or from storage device 408
Program in memory (RAM) 403 and execute various movements appropriate and processing.In RAM 403, it is also stored with electronic equipment
Various programs and data needed for 400 operations.Processing unit 401, ROM 402 and RAM 403 pass through bus or communication line
404 are connected with each other.Input/output (I/O) interface 405 is also connected to bus or communication line 404.
In general, following device can connect to I/O interface 405: including such as touch screen, touch tablet, keyboard, mouse, figure
As the input unit 406 of sensor, microphone, accelerometer, gyroscope etc.;Including such as liquid crystal display (LCD), loudspeaking
The output device 407 of device, vibrator etc.;Storage device 408 including such as tape, hard disk etc.;And communication device 409.It is logical
T unit 409 can permit electronic equipment 400 and wirelessly or non-wirelessly be communicated with other equipment to exchange data.Although Fig. 4 shows
The electronic equipment 400 with various devices is gone out, it should be understood that being not required for implementing or having all dresses shown
It sets.It can alternatively implement or have more or fewer devices.
Particularly, in accordance with an embodiment of the present disclosure, it may be implemented as computer above with reference to the process of flow chart description
Software program.For example, embodiment of the disclosure includes a kind of computer program product comprising be carried on computer-readable medium
On computer program, which includes the program code for method shown in execution flow chart.In such reality
It applies in example, which can be downloaded and installed from network by communication device 409, or from storage device 408
It is mounted, or is mounted from ROM 402.When the computer program is executed by processing unit 401, the embodiment of the present disclosure is executed
Method in the above-mentioned function that limits.
It should be noted that the above-mentioned computer-readable medium of the disclosure can be computer-readable signal media or meter
Calculation machine readable storage medium storing program for executing either the two any combination.Computer readable storage medium for example including but be not limited to
Electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor system, device or device, or any above combination.It is computer-readable
The more specific example of storage medium can include but is not limited to: have electrical connection, the portable computing of one or more conducting wires
Machine disk, hard disk, random access storage device (RAM), read-only memory (ROM), 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 the disclosure, computer readable storage medium can be it is any include or storage program it is tangible
Medium, the program can be commanded execution system, device or device use or in connection.And in the disclosure,
Computer-readable signal media may include in a base band or as the data-signal that carrier wave a part is propagated, wherein carrying
Computer-readable program code.The data-signal of this propagation can take various forms, including but not limited to electromagnetic signal,
Optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be other than computer readable storage medium
Any computer-readable medium, which can send, propagates or transmit for by instruction execution
System, device or device use or program in connection.The program code for including on computer-readable medium can
To transmit with any suitable medium, including but not limited to: electric wire, optical cable, RF (radio frequency) etc. are above-mentioned any appropriate
Combination.
Above-mentioned computer-readable medium can be included in above-mentioned electronic equipment;It is also possible to individualism, and not
It is fitted into the electronic equipment.
Above-mentioned computer-readable medium carries one or more program, when said one or multiple programs are by the electricity
When sub- equipment executes, so that the method that the electronic equipment executes the training convolutional neural networks classifier in above-described embodiment.
The calculating of the operation for executing the disclosure can be write with one or more programming languages or combinations thereof
Machine program code, above procedure design language include object oriented program language-such as Java, Smalltalk, C++,
It further include conventional procedural programming language-such as " C " language or similar programming language.Program code can be complete
It executes, partly executed on the user computer on the user computer entirely, being executed as an independent software package, part
Part executes on the remote computer or executes on a remote computer or server completely on the user computer.It is relating to
And in the situation of remote computer, remote computer can include local area network (LAN) or wide area network by the network-of any kind
(WAN)-it is connected to subscriber computer, or, it may be connected to outer computer (such as led to using ISP
Cross internet connection).
Flow chart and block diagram in attached drawing are illustrated according to the system of the various embodiments of the disclosure, method and computer journey
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
A part of one module, program segment or code of table, a part of the module, program segment or code include one or more use
The executable instruction of the logic function as defined in realizing.It should also be noted that in some implementations as replacements, being marked in box
The function of note can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are actually
It can be basically executed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.Also it to infuse
Meaning, the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart can be with holding
The dedicated hardware based system of functions or operations as defined in row is realized, or can use specialized hardware and computer instruction
Combination realize.
Being described in unit involved in the embodiment of the present disclosure can be realized by way of software, can also be by hard
The mode of part is realized.Wherein, the title of unit does not constitute the restriction to the unit itself under certain conditions.
Above description is only the preferred embodiment of the disclosure and the explanation to institute's application technology principle.Those skilled in the art
Member is it should be appreciated that the open scope involved in the disclosure, however it is not limited to technology made of the specific combination of above-mentioned technical characteristic
Scheme, while should also cover in the case where not departing from design disclosed above, it is carried out by above-mentioned technical characteristic or its equivalent feature
Any combination and the other technical solutions formed.Such as features described above has similar function with (but being not limited to) disclosed in the disclosure
Can technical characteristic replaced mutually and the technical solution that is formed.
Claims (11)
1. a kind of method of trained classifier characterized by comprising
ID Card Image set is obtained, the ID Card Image in the ID Card Image set is corresponding with Chinese character label information, institute
It states Chinese character label information and is used to indicate the Chinese included by the predeterminable area of ID Card Image corresponding with the Chinese character label information
Word;
Determine the output project of convolutional neural networks classifier, the output project of the convolutional neural networks classifier and the Chinese
Word label information is corresponding;
According to the ID Card Image set training convolutional neural networks classifier.
2. the method for trained classifier according to claim 1, which is characterized in that the ID Card Image set includes the
One subclass and second subset are closed, indicated by the corresponding Chinese character label information of ID Card Image in first subclass
The Chinese character include Chinese characters in common use, the second subset close in the corresponding Chinese character label information of ID Card Image it is signified
The Chinese character shown includes rare Chinese character.
3. the method for trained classifier according to claim 2, which is characterized in that instructed according to the ID Card Image set
Practice the convolutional neural networks classifier, comprising:
The ID Card Image of the first quantity is determined from first subclass;
The ID Card Image of the second quantity is determined from second subset conjunction;
The convolutional Neural net is updated according to the ID Card Image of first quantity and the ID Card Image of second quantity
The parameter of network classifier.
4. the method for trained classifier according to claim 3, which is characterized in that first quantity is 96, described the
Two quantity are 32.
5. according to the method for the training classifier any in claim 2-4, which is characterized in that the acquisition identity card figure
Image set closes, including obtains the second subset and close;
The second subset is obtained to close, comprising:
Obtain identity card template image;
Obtain rare Chinese character image;
The ID Card Image in the second subset conjunction is synthesized according to the rare Chinese character image and the identity card template image.
6. the method for trained classifier according to claim 5, which is characterized in that the acquisition rare Chinese character image, packet
It includes:
The rare Chinese character image is obtained based on No. 5 words of the Song typeface, the color of the rare Chinese character in the rare Chinese character image is black
Color, the background color of the rare Chinese character image are white;
The identity card synthesized according to the rare Chinese character image and the identity card template image in the second subset conjunction
Image, comprising:
The rare Chinese character image is covered to the predeterminable area of the identity card template image.
7. the method for trained classifier according to claim 6, which is characterized in that the identity card template image it is described
Predeterminable area includes identity card shading.
8. the method for trained classifier according to claim 7, which is characterized in that by the rare Chinese character image cover to
The predeterminable area of the identity card template image, comprising:
The rare Chinese character of black in the rare Chinese character image is covered to the preset areas of the identity card template image
Domain.
9. a kind of device of trained classifier characterized by comprising
Image collection obtains module, the ID Card Image for obtaining ID Card Image set, in the ID Card Image set
Corresponding with classification information, the classification information indicates the classification of ID Card Image corresponding with the classification information;
Determining module, for determining the output project of convolutional neural networks classifier, the output project and the classification information
It is corresponding;
Training module, for according to the ID Card Image set and convolutional neural networks (CNN) the training convolutional Neural net
Network classifier.
10. a kind of electronic equipment, comprising:
Memory, for storing computer-readable instruction;And
Processor, for running the computer-readable instruction, so that realizing according to claim 1-8 when the processor is run
Any one of described in training classifier method.
11. a kind of non-transient computer readable storage medium, for storing computer-readable instruction, when the computer-readable finger
When order is executed by computer, so that the computer perform claim requires the side of training classifier described in any one of 1-8
Method.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910439084.1A CN110222746A (en) | 2019-05-24 | 2019-05-24 | Method, apparatus, electronic equipment and the computer readable storage medium of training classifier |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910439084.1A CN110222746A (en) | 2019-05-24 | 2019-05-24 | Method, apparatus, electronic equipment and the computer readable storage medium of training classifier |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110222746A true CN110222746A (en) | 2019-09-10 |
Family
ID=67818116
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910439084.1A Pending CN110222746A (en) | 2019-05-24 | 2019-05-24 | Method, apparatus, electronic equipment and the computer readable storage medium of training classifier |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110222746A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111798250A (en) * | 2019-12-31 | 2020-10-20 | 支付宝实验室(新加坡)有限公司 | Method and system for determining the authenticity of official documents |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102163287A (en) * | 2011-03-28 | 2011-08-24 | 北京邮电大学 | Method for recognizing characters of licence plate based on Haar-like feature and support vector machine |
CN105512657A (en) * | 2015-08-20 | 2016-04-20 | 北京旷视科技有限公司 | Character recognition method and apparatus |
CN106325750A (en) * | 2016-08-26 | 2017-01-11 | 曹蕊 | Character recognition method and system applied in terminal equipment |
CN108805131A (en) * | 2018-05-22 | 2018-11-13 | 北京旷视科技有限公司 | Text line detection method, apparatus and system |
-
2019
- 2019-05-24 CN CN201910439084.1A patent/CN110222746A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102163287A (en) * | 2011-03-28 | 2011-08-24 | 北京邮电大学 | Method for recognizing characters of licence plate based on Haar-like feature and support vector machine |
CN105512657A (en) * | 2015-08-20 | 2016-04-20 | 北京旷视科技有限公司 | Character recognition method and apparatus |
CN106325750A (en) * | 2016-08-26 | 2017-01-11 | 曹蕊 | Character recognition method and system applied in terminal equipment |
CN108805131A (en) * | 2018-05-22 | 2018-11-13 | 北京旷视科技有限公司 | Text line detection method, apparatus and system |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111798250A (en) * | 2019-12-31 | 2020-10-20 | 支付宝实验室(新加坡)有限公司 | Method and system for determining the authenticity of official documents |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107633218A (en) | Method and apparatus for generating image | |
CN110458918A (en) | Method and apparatus for output information | |
CN107644209A (en) | Method for detecting human face and device | |
CN110288082A (en) | Convolutional neural networks model training method, device and computer readable storage medium | |
CN110321958A (en) | Training method, the video similarity of neural network model determine method | |
CN108846440A (en) | Image processing method and device, computer-readable medium and electronic equipment | |
CN108510472A (en) | Method and apparatus for handling image | |
CN109086719A (en) | Method and apparatus for output data | |
CN110047124A (en) | Method, apparatus, electronic equipment and the computer readable storage medium of render video | |
CN109829432A (en) | Method and apparatus for generating information | |
CN110532983A (en) | Method for processing video frequency, device, medium and equipment | |
CN109961032A (en) | Method and apparatus for generating disaggregated model | |
CN106489141A (en) | For method that content is classified and electronic equipment | |
CN109815365A (en) | Method and apparatus for handling video | |
CN109948699A (en) | Method and apparatus for generating characteristic pattern | |
CN108491812A (en) | The generation method and device of human face recognition model | |
CN110062157A (en) | Render method, apparatus, electronic equipment and the computer readable storage medium of image | |
CN108509994A (en) | character image clustering method and device | |
CN108446658A (en) | The method and apparatus of facial image for identification | |
CN110211195A (en) | Generate method, apparatus, electronic equipment and the computer readable storage medium of image collection | |
CN109948700A (en) | Method and apparatus for generating characteristic pattern | |
CN109285181A (en) | The method and apparatus of image for identification | |
CN110381352A (en) | Display methods, device, electronic equipment and the readable medium of virtual present | |
CN109977832A (en) | A kind of image processing method, device and storage medium | |
CN110516099A (en) | Image processing method and device |
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 |