CN107423700A - The method and device of testimony verification - Google Patents
The method and device of testimony verification Download PDFInfo
- Publication number
- CN107423700A CN107423700A CN201710581244.7A CN201710581244A CN107423700A CN 107423700 A CN107423700 A CN 107423700A CN 201710581244 A CN201710581244 A CN 201710581244A CN 107423700 A CN107423700 A CN 107423700A
- Authority
- CN
- China
- Prior art keywords
- network
- facial image
- certificate
- msub
- sample
- 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
- 238000012795 verification Methods 0.000 title claims abstract description 63
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000001815 facial effect Effects 0.000 claims abstract description 164
- 238000012549 training Methods 0.000 claims abstract description 68
- 230000006870 function Effects 0.000 claims description 42
- 238000009826 distribution Methods 0.000 claims description 9
- 238000003860 storage Methods 0.000 claims description 6
- 238000005286 illumination Methods 0.000 claims description 5
- 230000008447 perception Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 description 7
- 230000004069 differentiation Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 208000009119 Giant Axonal Neuropathy Diseases 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 201000003382 giant axonal neuropathy 1 Diseases 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Human Computer Interaction (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to the method and device of testimony verification.Methods described includes：The certificate facial image in certificate photo is obtained, gathers the natural light facial image of user；The good generation of certificate facial image input training in advance is resisted into network, facial image is rebuild according to corresponding to the output of the generation confrontation network obtains the certificate facial image；Wherein, the generation confrontation network is used to add the certificate facial image of input default natural light attribute information, and the high resolution of the reconstruction facial image of its output is in the resolution ratio of the certificate facial image；The reconstruction facial image and the natural light facial image are compared, testimony verification is carried out according to comparison result.The present invention can effectively improve the degree of accuracy of certification verification.
Description
Technical field
The present invention relates to technical field of face recognition, more particularly to the method, apparatus of testimony verification and other.
Background technology
With the fast development of face recognition technology, testimony verification system also obtains extensive concern, real in security protection, bank etc.
The demand of border application scenarios increases severely.In isomery face recognition technology, the difference of mode makes facial image difference huge, is to cause
The main reason for being difficult to accurately differentiate.Testimony verification problem belongs to isomery recognition of face, refers to the certificate for judging low resolution
Whether illumination face image matches with the natural lighting facial image of high-resolution.
The greatest problem of testimony verification system is that the degree of accuracy of verification is relatively low, and based on traditional isomery recognition of face side
Method, the otherness for being only capable of eliminating mode in terms of feature extraction and measuring similarity two are examined, can not be preferably applicable
In testimony verification.
The content of the invention
Based on this, the embodiment of the present invention provides the method and device of testimony verification, can effectively solve the problem that mode isomery causes
Testimony verification accuracy it is low the problem of.
The method that one aspect of the present invention provides testimony verification, including：
The certificate facial image in certificate photo is obtained, gathers the natural light facial image of user；
The good generation of certificate facial image input training in advance is resisted into network, network is resisted according to the generation
Output obtains rebuilding facial image corresponding to the certificate facial image；Wherein, the generation confrontation network is used for input
Certificate facial image adds default natural light attribute information, and the high resolution of the reconstruction facial image of its output is in described
The resolution ratio of certificate facial image；
The reconstruction facial image and the natural light facial image are compared, testimony verification is carried out according to comparison result.
One aspect of the present invention provides a kind of device of testimony verification, including：
Man face image acquiring module, for obtaining the certificate facial image in certificate photo, gather the natural light face of user
Image；
Face image module, network is resisted for the certificate facial image to be inputted into the good generation of training in advance,
Facial image is rebuild according to corresponding to the output of the generation confrontation network obtains the certificate facial image；Wherein, the life
It is used to add the certificate facial image of input default natural light attribute information, and the reconstruction people of its output into confrontation network
The high resolution of face image is in the resolution ratio of the certificate facial image；
Testimony verification module, for comparing the reconstruction facial image and the natural light facial image, tied according to comparing
Fruit carries out testimony verification.
One aspect of the present invention provides a kind of computer equipment, including memory, processor and storage are on a memory and can
The computer program run on a processor, the step of realizing method described above during the computing device described program.
Above-mentioned technical proposal, the certificate facial image in certificate photo is obtained, gather the natural light facial image of user；Pass through
The good generation of certificate facial image input training in advance is resisted into network, obtained according to the output that the generation resists network
Facial image is rebuild corresponding to the certificate facial image；Wherein, the generation confrontation network is used for the certificate face to input
Image adds default natural light attribute information, and the high resolution of the reconstruction facial image of its output is in the certificate face
The resolution ratio of image；Testimony verification is carried out finally by the comparison reconstruction facial image and the natural light facial image.By
This can realize the super-resolution rebuilding of certificate illumination face image (by low point on the premise of retaining original certificate and shining image information
Resolution image is converted into high-definition picture), certificate photo is converted to the mode of natural lighting, and then isomery testimony verification is asked
Topic is converted into general recognition of face problem, effectively increases the degree of accuracy of testimony verification.
Brief description of the drawings
Fig. 1 is the indicative flowchart of the method for the testimony verification of an embodiment；
Fig. 2 is the structural representation of the maker network of an embodiment；
Fig. 3 is the indicative flowchart of the generation confrontation network retraining of an embodiment；
Fig. 4 is the schematic diagram of the device of the testimony verification of an embodiment.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Fig. 1 is the indicative flowchart of the method for the testimony verification of an embodiment；As shown in figure 1, the people in the present embodiment
The method that card is examined includes step：
S11, the certificate facial image in certificate photo is obtained, gather the natural light facial image of user；
It should be understood that the certificate being related in the embodiment of the present invention can be any certificate for being accompanied with facial image, such as：
Identity card, passport, driver's license, the pass or enterprise's work card etc..The certificate photo can be pasted onto additional clause photo,
Can be the photo being printed on certificate, such as China second-generation identity card.In addition, the certificate photo both can be black and white, can also
It is colored.
Need what is illustrated, the natural light facial image, refer to the image gathered in real time in current environment, both included
The image gathered in real time under outdoor natural light environment, the also image including being gathered in real time under Interior Illumination Environment.
S12, the certificate facial image is inputted into the good generation of training in advance and resists network, net is resisted according to the generation
The output of network obtains rebuilding facial image corresponding to the certificate facial image；
Generation confrontation network (Generative Adversarial Networks, abbreviation GANs), it is a kind of generation mould
Type, its basic thought is to obtain many training samples in training storehouse, so as to learn the probability distribution of these training cases generation.
Its implementation method is to allow two network models to vie each other, and carries out a game.One of them is called maker network
(or maker, generation network), seem the image of ' nature ' for generating, it is desirable to initial data distribution as far as possible one
Cause；Another is called arbiter network (or arbiter, differentiate network), for judge whether given image seems ' from
So ', in other words, if seem that artificial (machine) generates.The target of maker network is that ' deceiving ' crosses arbiter network, is differentiated
The target of device network is not working as maker network.When two groups of network models are constantly trained, maker network is continuously generated newly
Result attempted, both abilities improve mutually, until maker network generation artificial sample seem and original sample
Originally it is not different.
The resolution ratio of the certificate facial image generally collected from certificate is relatively low, and net is resisted by the generation of this step
Network can carry out super-resolution reconstruction to the certificate facial image of low point of ratio, add default natural light attribute information, can also mend
The pixel portion lacked in neat original image.
In embodiments of the present invention, the high resolution of the reconstruction facial image of the generation confrontation network output is in the card
The resolution ratio of part facial image.Ideally, it is described generation confrontation network obtain reconstruction facial image mode with it is corresponding
Natural light facial image mode it is consistent.
S13, the reconstruction facial image and the natural light facial image are compared, testimony of a witness core is carried out according to comparison result
It is real.
Based on above-mentioned steps, certificate facial image is rebuild (i.e. mode conversion), therefore step S13 is
Common facial image identification, rather than the identification of isomery facial image, the complexity of identification is on the one hand reduced, on the other hand,
Be advantageous to improve the degree of accuracy of identification.
, can be on the premise of retaining original certificate and shining image information by the method for the testimony verification of above-described embodiment, will
Certificate photo is converted to the mode of natural lighting, and then isomery testimony verification problem is converted into the identification of in general facial image and asked
Topic, effectively increases the degree of accuracy of testimony verification.
In an alternative embodiment, include in above-mentioned steps S11：The certificate photo gathered in evidence carries out face inspection afterwards
Survey, alignment, cut out the human face region in human face five-sense-organ region, obtain certificate facial image.The natural light facial image is preferred
For the image of positive high definition.
In an alternative embodiment, the generation confrontation network used in above-mentioned steps S12 includes maker network and differentiation
Device network.Wherein, the maker network includes 6 layers of residual error convolutional network structure, as shown in Fig. 2 wherein preceding 3 layers are convolution
Layer, latter 3 layers are reverse convolutional layer, and facial image is rebuild in last reverse convolutional layer output.Specific network is formed for example：Institute
State in 6 layers of residual error convolutional network structure, upper batch normalization layers (abbreviation BN is connected after each 3 × 3 convolutional layer
Layer) and ReLU activation primitive layers so that the input of each convolutional layer keeps same distribution, and network entirety training speed is fast.
It is set to 1, pad from every layer of first five layer feature of extraction 64 maps, stride and is set to 0, last reverse convolutional layer is used for
Reconstruction image.This network structure can not only make network be easier to train, while retain and utilize the certificate facial image inputted
Reconstruction for the image information high-definition picture.Preferably, the pixel size and step S11 of the reconstruction image of maker network output
The pixel size of the natural light facial image of middle collection is consistent.The arbiter network includes Light CNN residual error network structures,
The network structure is advantageous to Enhanced feature robustness and reduces network parameter.Preferably, using the Network in Light CNN
In Network network structures, activation primitive is used as by the use of Max Feature Map operations.Using Network in Network
The advantages of network structure, includes：Preferably local abstract, smaller global Overfitting and less network parameter.
In an alternative embodiment, in addition to the step of training in advance generation confrontation network, including to maker network
Training and the training to arbiter network.The mode of training can be：It is primarily based on ImageNet databases and network is resisted to generation
Carry out pre-training；Then retraining is carried out to the generation confrontation network Jing Guo pre-training based on default testimony of a witness Sample Storehouse again, directly
Network is resisted to the generation for being met preparatory condition.Wherein, ImageNet databases are that image recognition is maximum in the world at present
Database；The testimony of a witness Sample Storehouse includes natural light face figure corresponding to multiple certificate photo samples and each certificate photo sample
Decent.By the training of two stage disparate databases, the generation confrontation network for meeting testimony verification demand can be obtained, with
The certificate facial image of low resolution is converted into high-resolution reconstruction facial image.
In an alternative embodiment, the generation confrontation network Jing Guo pre-training is carried out again based on default testimony of a witness Sample Storehouse
Training includes：The maker network training in network is resisted to generation and the arbiter network training in network is resisted to generation,
Two training process are influenced each other so that two networks are gradually improved to the network model for meeting testimony verification demand.
Wherein, the process of the maker network training in network is resisted to generation to be included：Card is obtained from testimony of a witness Sample Storehouse
Part this and its corresponding natural light facial image sample in the same old way, obtain certificate facial image sample from the certificate photo sample,
Input using certificate facial image sample as maker network, the net of the maker network is trained based on quadratic loss function
Network parameter, until the quadratic loss function minimizes；The quadratic loss function is natural light facial image sample and generation
The function of the difference of two squares based on pixel of the reconstruction facial image of device network output.
Wherein, the process of the arbiter network training in network is resisted to generation to be included：By the natural light facial image
The input of sample and the reconstruction facial image of maker network output as arbiter network, is trained based on loss function is perceived
The network parameter of the network parameter of the arbiter network and the maker network；The perception loss function is by maker
The reconstruction facial image of network output is determined as the function of the probability of true nature light facial image.
In a preferred embodiment, the training process of generation confrontation network is referred to shown in Fig. 3.Assuming that IyFor high-resolution
Natural light facial image sample, IxFor the certificate facial image sample of low resolution, IsTo rebuild facial image；Wherein rebuild
Facial image is identical with the pixel size of natural light facial image sample.Generation confrontation network is by maker networkAnd differentiation
Device networkComposition, θ represent network parameter to be trained in generation confrontation network.
To the certificate facial image sample I of maker network inputs low resolutionx, the reconstruction people of maker network output
Face image IsTo maker network.Function model corresponding to maker network isIn an alternative embodiment, described in training
The object function of the network parameter of maker networkFor：
θGRepresent the network parameter of maker network, lsFor quadratic loss function, N is to participate in training the total of certificate photo sample
Number, IyRepresent natural light facial image sample, IxRepresent certificate facial image sample, IsRepresent that certificate facial image sample is corresponding
Reconstruction facial image.
Preferably, the loss function of the Squared Error Loss (MSE) based on pixel is used in the maker network training,
I.e.：
Wherein, r represents the size ratio of natural light facial image and certificate facial image；W represents that certificate facial image exists
The pixel of width, H represent pixel of the certificate facial image in length direction.
Maker network is exported and rebuilds facial image IsWith high-resolution natural light facial image sample IyInput differentiates
Device network, arbiter network is trained.Function model corresponding to the arbiter network is expressed asθDRepresent to differentiate
The network parameter of device network, its task are to judge maker networkCaused reconstruction image(i.e. Is) whether be it is true,
Solves " minimax game " problem, object function is represented by：
Wherein, Ε represents mathematic expectaion, Iy~pdata(Iy) represent natural light facial image sample IyMeet high resolution graphics
The probability distribution of picture is pdata(Iy)；Ix~pG(Ix) represent certificate facial image sample IxThe probability distribution for meeting maker is pG
(Iy)；Log represents logarithm operation；It is by natural light facial image sample IyIt is determined as true nature light facial image
Probability；Represent arbiter networkBy maker networkThe reconstruction image of outputIt is determined as
The probability of true nature light facial image.
Train arbiterMaximize input example and generate the probability of the correct label of sample, while train makerMinimizeWhen global optimum, there is pdata=pG, i.e. maker network can be fitted high score completely
The probability distribution of resolution image.
Preferably, if the reconstruction image that arbiter network exports maker networkIt is determined as true nature light people
The probability of face image, then arbiter network output 1, otherwise, output -1.According to the differentiation result optimizing arbiter of arbiter network
Network.The object function for optimizing arbiter network is the related function of the differentiation result of arbiter network.
Because the MSE loss functions that maker network training uses can lose the high-frequency information of input picture to a certain degree,
Blooming is caused, therefore is added in network is differentiated and perceives loss function, from the orientation optimization arbiter network of perception.Adopt
With perception loss functionOptimize the object function of arbiter network：
Wherein,Represent arbiter networkBy maker networkThe reconstruction image of output
It is determined as the probability of true nature light facial image.
When maker networkWith arbiter networkWhen training is completed, the training knot of the generation confrontation network
Beam.When carrying out testimony verification, the original certificate facial image of collection can be carried out by weight based on the generation confrontation network trained
Build, and then be authenticated examining based on the natural light facial image rebuild facial image and gathered in real time, improve the accurate of verification
Degree.
It can be seen that the task of whole generation confrontation network will make arbiter network just as a game, maker network
Obscure the true or false of reconstruction image, and the target of arbiter network is the authenticity of resolution image as far as possible.Therefore to generation
The training method for resisting network is different from the conventional mode for minimizing pixel error, and thus obtained generation confrontation network can not only
The high-frequency information of the original image of input is effectively retained, while the reconstruction of high similarity can also be produced by way of sensing and optimizing
Image.
The method of the testimony verification of above-described embodiment can effectively solve the image Heterogeneity in testimony verification, with two generation bodies
Exemplified by the testimony verification of part card, it can obtain resisting net for the generation of China second-generation identity card super-resolution rebuilding by depth training
Network, in the case where retaining original identity document according to information compensation natural light attribute information (such as the shading value in each region of face,
Illumination and color etc.), export high-resolution China second-generation identity card human face rebuilding image.Then with the natural lighting face figure of collection
As being compared, using existing face recognition technology, the testimony verification of China second-generation identity card can be effectively carried out.
It should be noted that for foregoing each method embodiment, in order to which simplicity describes, it is all expressed as a series of
Combination of actions, but those skilled in the art should know, the present invention is not limited by described sequence of movement, because according to
According to the present invention, some steps can use other orders or carry out simultaneously.
Based on the method identical thought with the testimony verification in above-described embodiment, the present invention also provides the dress of testimony verification
Put, the device can be used for the method for performing above-mentioned testimony verification.For convenience of description, the structure of the device embodiment of testimony verification
In schematic diagram, the part related to the embodiment of the present invention is illustrate only, it will be understood by those skilled in the art that schematic structure is simultaneously
The not restriction of structure twin installation, it can include than illustrating more or less parts, either combine some parts or different
Part is arranged.
Fig. 4 is the schematic diagram of the device of the testimony verification of one embodiment of the invention, as shown in figure 4, the present embodiment
The device of testimony verification include：Man face image acquiring module 410, face image module 420 and testimony verification module
430, details are as follows for each module：
The man face image acquiring module 410, for obtaining the certificate facial image in certificate photo, gather the nature of user
Light facial image
The face image module 420, for the certificate facial image to be inputted into the good generation pair of training in advance
Anti- network, facial image is rebuild according to corresponding to the output of the generation confrontation network obtains the certificate facial image；Wherein,
The generation confrontation network is used to add the certificate facial image of input default natural light attribute information, and its output
The high resolution of facial image is rebuild in the resolution ratio of the certificate facial image.
The testimony verification module 430, for comparing the reconstruction facial image and the natural light facial image, according to
Comparison result carries out testimony verification.
In an alternative embodiment, in addition to network training module, generate confrontation network for training.The network training mould
Block is specifically used for, and pre-training is carried out to generation confrontation network based on ImageNet databases；Based on default testimony of a witness Sample Storehouse pair
Generation confrontation network by pre-training carries out retraining, until the generation for being met preparatory condition resists network.
Preferably, the generation confrontation network includes maker network and arbiter network；The maker network includes 6
Layer residual error convolutional network structure, wherein first 3 layers are convolutional layer, latter 3 layers are reverse convolutional layer, last reverse convolutional layer output
Rebuild facial image.The arbiter network includes Light CNN residual error network structures.The network structure can effectively be accelerated to generate
The training process of network is resisted, shortens the training time.
In an alternative embodiment, the network training module includes：First training unit and the second training unit.
First training unit, for resisting the maker network training in network, specific training method bag to generation
Include：Certificate photo sample and its corresponding natural light facial image sample are obtained from testimony of a witness Sample Storehouse, using certificate photo sample as
The input of maker network, the network parameter of the maker network is trained based on quadratic loss function；The Squared Error Loss letter
The function of the difference of two squares based on pixel for the reconstruction facial image that number exports for natural light facial image sample with maker network.
Second training unit, for resisting the arbiter network training in network, specific training method bag to generation
Include：The reconstruction facial image that the natural light facial image sample and maker network are exported is as the defeated of arbiter network
Enter, the network parameter of the arbiter network and the network parameter of the maker network are trained based on loss function is perceived；Institute
It is the probability that the reconstruction facial image that maker network exports is determined as to true high-definition picture to state and perceive loss function
Function.
Preferably, the training process of generation confrontation network refers to detailed process shown in Fig. 3 and can refer to above method implementation
Described in example.
It should be noted that in the embodiment of the device of the testimony verification of above-mentioned example, the letter between each module/unit
The contents such as interaction, implementation procedure are ceased, due to being based on same design, its technique effect brought with preceding method embodiment of the present invention
Identical with preceding method embodiment of the present invention, particular content can be found in the narration in the inventive method embodiment, no longer superfluous herein
State.
In addition, in the embodiment of the device of the testimony verification of above-mentioned example, the logical partitioning of each functional module is only to lift
Example explanation, can be as needed in practical application, for example, for corresponding hardware configuration requirement or software realization facility
Consider, above-mentioned function distribution completed by different functional module, will the testimony verification device internal structure division
Into different functional modules, to complete all or part of function described above.Wherein each function mould can both use hardware
Form realize, can also be realized in the form of software function module.
It will appreciated by the skilled person that realizing all or part of flow in above-described embodiment method, being can
To instruct the hardware of correlation to complete by computer program, described program can be stored in a computer-readable storage and be situated between
In matter, as independent production marketing or use.Described program upon execution, can perform the complete of such as embodiment of above-mentioned each method
Portion or part steps.In addition, the storage medium is also settable with a kind of computer equipment, also being wrapped in the computer equipment
Include processor, during program in storage medium described in the computing device, can realize above-mentioned each method embodiment it is complete
Portion or part steps.Wherein, described storage medium can be magnetic disc, CD, read-only memory (Read-Only
Memory, ROM) or random access memory (Random Access Memory, RAM) etc..
In the above-described embodiments, the description to each embodiment all emphasizes particularly on different fields, and does not have the portion being described in detail in some embodiment
Point, it may refer to the associated description of other embodiments.
Embodiment described above only expresses the several embodiments of the present invention, it is impossible to is interpreted as to the scope of the claims of the present invention
Limitation.It should be pointed out that for the person of ordinary skill of the art, without departing from the inventive concept of the premise,
Various modifications and improvements can be made, these belong to protection scope of the present invention.Therefore, the protection domain of patent of the present invention
It should be determined by the appended claims.
Claims (11)
- A kind of 1. method of testimony verification, it is characterised in that including：The certificate facial image in certificate photo is obtained, gathers the natural light facial image of user；The good generation of certificate facial image input training in advance is resisted into network, the output of network is resisted according to the generation Obtain rebuilding facial image corresponding to the certificate facial image；Wherein, the generation confrontation network is used for the certificate to input Facial image adds default natural light attribute information, and the high resolution of the reconstruction facial image of its output is in the certificate The resolution ratio of facial image；The reconstruction facial image and the natural light facial image are compared, testimony verification is carried out according to comparison result.
- 2. the method for testimony verification according to claim 1, it is characterised in that also include：Training generation confrontation network Step, the step include：Pre-training is carried out to generation confrontation network based on ImageNet databases；Retraining is carried out to the generation confrontation network Jing Guo pre-training based on default testimony of a witness Sample Storehouse；Wherein, the testimony of a witness sample This storehouse includes natural light facial image sample corresponding to multiple certificate photo samples and each certificate photo sample.
- 3. the method for testimony verification according to claim 2, it is characterised in that the generation confrontation network includes maker Network and arbiter network；The maker network includes 6 layers of residual error convolutional network structure, wherein first 3 layers are convolutional layer, latter 3 layers are reverse convolution Facial image is rebuild in layer, last reverse convolutional layer output；The arbiter network includes Light CNN residual error network structures.
- 4. the method for testimony verification according to claim 2, it is characterised in that described to be based on default testimony of a witness Sample Storehouse pair Generation confrontation network by pre-training carries out retraining, including：The maker network training in network is resisted to generation based on default testimony of a witness Sample Storehouse, is specifically included：Certificate photo sample and its corresponding natural light facial image sample are obtained from testimony of a witness Sample Storehouse, from the certificate photo sample In obtain certificate facial image sample, the input using certificate facial image sample as maker network, based on Squared Error Loss letter Number trains the network parameter of the maker network；The quadratic loss function is natural light facial image sample and maker net The function of the difference of two squares based on pixel of the reconstruction facial image of network output；The arbiter network training in network is resisted to generation based on default testimony of a witness Sample Storehouse, is specifically included：The reconstruction facial image that the natural light facial image sample and maker network are exported is as arbiter network Input, the network parameter of the arbiter network and the network parameter of the maker network are trained based on loss function is perceived； The perception loss function is that the reconstruction facial image that maker network exports is determined as true nature light people by arbiter network The function of the probability of face image.
- 5. the method for testimony verification according to claim 4, it is characterised in that Function Modules corresponding to the maker network Type isTrain the object function of the network parameter of the maker networkFor：<mrow> <msub> <mover> <mi>&theta;</mi> <mo>^</mo> </mover> <mi>G</mi> </msub> <mo>=</mo> <mi>arg</mi> <munder> <mi>min</mi> <msub> <mi>&theta;</mi> <mi>G</mi> </msub> </munder> <mfrac> <mn>1</mn> <mi>N</mi> </mfrac> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>n</mi> <mo>-</mo> <mn>1</mn> </mrow> <mi>N</mi> </msubsup> <msup> <mi>l</mi> <mi>s</mi> </msup> <mrow> <mo>(</mo> <msub> <mi>G</mi> <msub> <mi>&theta;</mi> <mi>G</mi> </msub> </msub> <mo>(</mo> <msubsup> <mi>I</mi> <mi>n</mi> <mi>x</mi> </msubsup> <mo>)</mo> <mo>,</mo> <msubsup> <mi>I</mi> <mi>n</mi> <mi>y</mi> </msubsup> <mo>)</mo> </mrow> </mrow>θ represents the network parameter of generation confrontation network, θGRepresent the network parameter of maker network, lsFor quadratic loss function, N To participate in the sum of training certificate photo sample, IyRepresent natural light facial image sample, IxRepresent certificate facial image sample, Is Represent certificate facial image sample IxCorresponding reconstruction facial image；And/orFunction model corresponding to the arbiter networkThe object function for training the arbiter network is：<mrow> <munder> <mi>min</mi> <msub> <mi>&theta;</mi> <mi>G</mi> </msub> </munder> <munder> <mi>max</mi> <msub> <mi>&theta;</mi> <mi>G</mi> </msub> </munder> <msub> <mi>E</mi> <mrow> <msup> <mi>I</mi> <mi>y</mi> </msup> <mo>~</mo> <msub> <mi>p</mi> <mrow> <mi>d</mi> <mi>a</mi> <mi>t</mi> <mi>a</mi> </mrow> </msub> <mrow> <mo>(</mo> <msup> <mi>I</mi> <mi>y</mi> </msup> <mo>)</mo> </mrow> </mrow> </msub> <mo>&lsqb;</mo> <mi>log</mi> <mi> </mi> <msub> <mi>D</mi> <msub> <mi>&theta;</mi> <mi>G</mi> </msub> </msub> <mrow> <mo>(</mo> <msup> <mi>I</mi> <mi>y</mi> </msup> <mo>)</mo> </mrow> <mo>&rsqb;</mo> <mo>+</mo> <msub> <mi>E</mi> <mrow> <msup> <mi>I</mi> <mi>x</mi> </msup> <mo>~</mo> <msub> <mi>p</mi> <mi>G</mi> </msub> <mrow> <mo>(</mo> <msup> <mi>I</mi> <mi>x</mi> </msup> <mo>)</mo> </mrow> </mrow> </msub> <mo>&lsqb;</mo> <mi>log</mi> <mo>(</mo> <mn>1</mn> <mo>-</mo> <msub> <mi>D</mi> <msub> <mi>&theta;</mi> <mi>G</mi> </msub> </msub> <mo>(</mo> <mrow> <msub> <mi>G</mi> <msub> <mi>&theta;</mi> <mi>G</mi> </msub> </msub> <mrow> <mo>(</mo> <msup> <mi>I</mi> <mi>x</mi> </msup> <mo>)</mo> </mrow> </mrow> <mo>)</mo> <mo>&rsqb;</mo> </mrow>Wherein, θDThe network parameter of arbiter network is represented, Ε is mathematic expectaion, Iy~pdata(Iy) represent natural light facial image Sample IyThe probability distribution for meeting high-definition picture is pdata(Iy)；Ix~pG(Ix) represent certificate facial image sample IxMeet The probability distribution of maker is pG(Iy)；Log represents logarithm operation；It is by natural light facial image sample IyDifferentiate For the probability of true nature light facial image,Represent arbiter networkBy maker networkOutput Reconstruction imageIt is determined as the probability of true nature light facial image.
- 6. the method for testimony verification according to any one of claims 1 to 5, it is characterised in that described to compare the reconstruction people Face image and the natural light facial image, testimony verification is carried out according to comparison result, including：If the matching degree of the reconstruction facial image and the natural light facial image is more than given threshold, it is judged as testimony verification Pass through；Otherwise, it is judged as that testimony verification fails.
- 7. the method for testimony verification according to claim 6, it is characterised in that the natural lighting attribute includes light and shade Degree, illumination and/or color.
- A kind of 8. device of testimony verification, it is characterised in that including：Man face image acquiring module, for obtaining the certificate facial image in certificate photo, gather the natural light facial image of user；Face image module, network is resisted for the certificate facial image to be inputted into the good generation of training in advance, according to The output of the generation confrontation network obtains rebuilding facial image corresponding to the certificate facial image；Wherein, the generation pair Anti- network is used to add the certificate facial image of input in default natural light attribute information, and the reconstruction face figure of its output The high resolution of picture is in the resolution ratio of the certificate facial image；Testimony verification module, for comparing the reconstruction facial image and the natural light facial image, entered according to comparison result Row testimony verification.
- 9. the device of testimony verification according to claim 8, it is characterised in that also including network training module, for base Pre-training is carried out to generation confrontation network in ImageNet databases；Based on default testimony of a witness Sample Storehouse to the life Jing Guo pre-training Retraining is carried out into confrontation network, until the generation for being met preparatory condition resists network；Wherein, in the testimony of a witness Sample Storehouse Including natural light facial image sample corresponding to multiple certificate photo samples and each certificate photo sample.
- 10. the device of testimony verification according to claim 9, it is characterised in that the network training module includes：First training unit, for resisting the maker network training in network to generation, specifically include from testimony of a witness sample Certificate photo sample and its corresponding natural light facial image sample are obtained in storehouse, certificate face is obtained from the certificate photo sample Image pattern, the input using certificate facial image sample as maker network, the generation is trained based on quadratic loss function The network parameter of device network；The quadratic loss function is natural light facial image sample and the reconstruction people of maker network output The function of the difference of two squares based on pixel of face image；Second training unit, for resisting the arbiter network training in network to generation, specifically include：By the nature The input of light facial image sample and the reconstruction facial image of maker network output as arbiter network, is damaged based on perceiving Lose function and train the network parameter of the arbiter network and the network parameter of the maker network；The perception loss function The reconstruction facial image that maker network exports is determined as to the letter of the probability of true nature light facial image for arbiter network Number.
- 11. a kind of computer equipment, including memory, processor and storage are on a memory and the meter that can run on a processor Calculation machine program, it is characterised in that the step of any methods described of claim 1 to 7 is realized during the computing device described program Suddenly.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710581244.7A CN107423700B (en) | 2017-07-17 | 2017-07-17 | Method and device for verifying testimony of a witness |
| PCT/CN2018/093784 WO2019015466A1 (en) | 2017-07-17 | 2018-06-29 | Method and apparatus for verifying person and certificate |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710581244.7A CN107423700B (en) | 2017-07-17 | 2017-07-17 | Method and device for verifying testimony of a witness |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107423700A true CN107423700A (en) | 2017-12-01 |
| CN107423700B CN107423700B (en) | 2020-10-20 |
Family
ID=60429889
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710581244.7A Active CN107423700B (en) | 2017-07-17 | 2017-07-17 | Method and device for verifying testimony of a witness |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN107423700B (en) |
| WO (1) | WO2019015466A1 (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108090905A (en) * | 2018-01-05 | 2018-05-29 | 清华大学 | The determination methods and system of producing line exception |
| CN108280413A (en) * | 2018-01-17 | 2018-07-13 | 百度在线网络技术（北京）有限公司 | Face identification method and device |
| CN108416326A (en) * | 2018-03-27 | 2018-08-17 | 百度在线网络技术（北京）有限公司 | Face identification method and device |
| WO2019015466A1 (en) * | 2017-07-17 | 2019-01-24 | 广州广电运通金融电子股份有限公司 | Method and apparatus for verifying person and certificate |
| CN109711364A (en) * | 2018-12-29 | 2019-05-03 | 成都视观天下科技有限公司 | A kind of facial image super-resolution reconstruction method, device and computer equipment |
| CN110163114A (en) * | 2019-04-25 | 2019-08-23 | 厦门瑞为信息技术有限公司 | A kind of facial angle and face method for analyzing ambiguity, system and computer equipment |
| CN110263603A (en) * | 2018-05-14 | 2019-09-20 | 桂林远望智能通信科技有限公司 | Face identification method and device based on center loss and residual error visual simulation network |
| CN110580682A (en) * | 2019-09-16 | 2019-12-17 | 电子科技大学 | Countermeasure network seismic data super-resolution reconstruction method based on optimization generation |
| CN111553208A (en) * | 2020-04-15 | 2020-08-18 | 上海携程国际旅行社有限公司 | Identity recognition method, system, device and medium based on image of people and certificate integration |
| WO2020199577A1 (en) * | 2019-03-29 | 2020-10-08 | 北京市商汤科技开发有限公司 | Method and device for living body detection, equipment, and storage medium |
| CN112508782A (en) * | 2020-09-10 | 2021-03-16 | 浙江大华技术股份有限公司 | Network model training method, face image super-resolution reconstruction method and equipment |
| US11250329B2 (en) | 2017-10-26 | 2022-02-15 | Nvidia Corporation | Progressive modification of generative adversarial neural networks |
| US11263525B2 (en) | 2017-10-26 | 2022-03-01 | Nvidia Corporation | Progressive modification of neural networks |
Families Citing this family (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107578017B (en) * | 2017-09-08 | 2020-11-17 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating image |
| CN111583096A (en) * | 2019-02-15 | 2020-08-25 | 北京京东乾石科技有限公司 | Picture processing method and device, electronic equipment and computer readable medium |
| CN110097609B (en) * | 2019-04-04 | 2022-11-29 | 浙江凌迪数字科技有限公司 | Sample domain-based refined embroidery texture migration method |
| CN110097543B (en) * | 2019-04-25 | 2023-01-13 | 东北大学 | Hot-rolled strip steel surface defect detection method based on generation type countermeasure network |
| CN110119746B (en) * | 2019-05-08 | 2021-11-30 | 北京市商汤科技开发有限公司 | Feature recognition method and device and computer readable storage medium |
| CN110288537A (en) * | 2019-05-20 | 2019-09-27 | 湖南大学 | Facial image complementing method based on the depth production confrontation network from attention |
| CN110457994B (en) * | 2019-06-26 | 2024-05-10 | 平安科技（深圳）有限公司 | Face image generation method and device, storage medium and computer equipment |
| CN110443746B (en) * | 2019-07-25 | 2023-07-14 | 创新先进技术有限公司 | Picture processing method and device based on generation countermeasure network and electronic equipment |
| CN110517195B (en) * | 2019-07-26 | 2022-12-06 | 西安电子科技大学 | Unsupervised SAR image denoising method |
| CN112330526B (en) * | 2019-08-05 | 2024-02-09 | 深圳Tcl新技术有限公司 | Training method of face conversion model, storage medium and terminal equipment |
| CN111062290B (en) * | 2019-12-10 | 2023-04-07 | 西北大学 | Method and device for constructing Chinese calligraphy style conversion model based on generation confrontation network |
| CN111161200A (en) * | 2019-12-22 | 2020-05-15 | 天津大学 | Human body posture migration method based on attention mechanism |
| CN111476749B (en) * | 2020-04-03 | 2023-02-28 | 陕西师范大学 | Face repairing method for generating confrontation network in guiding mode based on face key points |
| CN111476717B (en) * | 2020-04-07 | 2023-03-24 | 西安电子科技大学 | Face image super-resolution reconstruction method based on self-attention generation countermeasure network |
| CN111597978B (en) * | 2020-05-14 | 2023-04-07 | 公安部第三研究所 | Method for automatically generating pedestrian re-identification picture based on StarGAN network model |
| CN111754478A (en) * | 2020-06-22 | 2020-10-09 | 怀光智能科技（武汉）有限公司 | Unsupervised domain adaptation system and unsupervised domain adaptation method based on generation countermeasure network |
| CN113761997B (en) * | 2020-08-27 | 2024-04-09 | 北京沃东天骏信息技术有限公司 | Method and device for generating semi-occlusion face recognition device |
| CN112070145B (en) * | 2020-09-04 | 2024-05-28 | 世纪易联(北京)科技有限公司 | Freshness attribute migration method of fruit image based on countermeasure network |
| CN112102186B (en) * | 2020-09-07 | 2024-04-05 | 河海大学 | Real-time enhancement method for underwater video image |
| CN112053303B (en) * | 2020-09-08 | 2024-04-05 | 河海大学 | Video image real-time enhancement method for underwater AUV |
| CN112233017B (en) * | 2020-10-28 | 2023-09-26 | 中国科学院合肥物质科学研究院 | Method for enhancing pathological face data based on generation countermeasure network |
| CN112598125B (en) * | 2020-11-25 | 2024-04-30 | 西安科技大学 | Handwriting digital generation method based on dual-discriminant weighting generation countermeasure network |
| CN114596236A (en) * | 2020-12-04 | 2022-06-07 | 国网智能科技股份有限公司 | Method and system for enhancing low-illumination image of closed cavity |
| CN112668623B (en) * | 2020-12-22 | 2024-04-16 | 中国铁道科学研究院集团有限公司 | Method and device for generating binaural pin defect sample based on generation countermeasure network |
| CN112818764B (en) * | 2021-01-15 | 2023-05-02 | 西安交通大学 | Low-resolution image facial expression recognition method based on feature reconstruction model |
| CN113034393A (en) * | 2021-03-25 | 2021-06-25 | 北京百度网讯科技有限公司 | Photo repairing method, device, equipment and storage medium |
| CN113378721B (en) * | 2021-06-11 | 2023-08-18 | 西安电子科技大学 | Symmetrical and local discrimination-based face correction method and system for generating countermeasure |
| CN113705400B (en) * | 2021-08-18 | 2023-08-15 | 中山大学 | Single-mode face living body detection method based on multi-mode face training |
| CN113762180B (en) * | 2021-09-13 | 2023-09-01 | 中国科学技术大学 | Training method and system for human body activity imaging based on millimeter wave radar signals |
| CN113780534B (en) * | 2021-09-24 | 2023-08-22 | 北京字跳网络技术有限公司 | Compression method, image generation method, device, equipment and medium of network model |
| CN114067399B (en) * | 2021-11-16 | 2024-03-15 | 桂林电子科技大学 | Face reconstruction and recognition method for non-matching scene |
| CN114913086B (en) * | 2022-05-05 | 2023-05-02 | 上海云思智慧信息技术有限公司 | Face image quality enhancement method based on generation countermeasure network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101694691A (en) * | 2009-07-07 | 2010-04-14 | 北京中星微电子有限公司 | Method and device for synthesizing facial images |
| WO2016112630A1 (en) * | 2015-01-12 | 2016-07-21 | 芋头科技（杭州）有限公司 | Image recognition system and method |
| CN106683048A (en) * | 2016-11-30 | 2017-05-17 | 浙江宇视科技有限公司 | Image super-resolution method and image super-resolution equipment |
| CN106845449A (en) * | 2017-02-22 | 2017-06-13 | 浙江维尔科技有限公司 | A kind of image processing apparatus, method and face identification system |
| CN106951867A (en) * | 2017-03-22 | 2017-07-14 | 成都擎天树科技有限公司 | Face identification method, device, system and equipment based on convolutional neural networks |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7127087B2 (en) * | 2000-03-27 | 2006-10-24 | Microsoft Corporation | Pose-invariant face recognition system and process |
| CN106845471A (en) * | 2017-02-20 | 2017-06-13 | 深圳市唯特视科技有限公司 | A kind of vision significance Forecasting Methodology based on generation confrontation network |
| CN106952229A (en) * | 2017-03-15 | 2017-07-14 | 桂林电子科技大学 | Image super-resolution rebuilding method based on the enhanced modified convolutional network of data |
| CN107423700B (en) * | 2017-07-17 | 2020-10-20 | 广州广电卓识智能科技有限公司 | Method and device for verifying testimony of a witness |
-
2017
- 2017-07-17 CN CN201710581244.7A patent/CN107423700B/en active Active
-
2018
- 2018-06-29 WO PCT/CN2018/093784 patent/WO2019015466A1/en active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101694691A (en) * | 2009-07-07 | 2010-04-14 | 北京中星微电子有限公司 | Method and device for synthesizing facial images |
| WO2016112630A1 (en) * | 2015-01-12 | 2016-07-21 | 芋头科技（杭州）有限公司 | Image recognition system and method |
| CN106683048A (en) * | 2016-11-30 | 2017-05-17 | 浙江宇视科技有限公司 | Image super-resolution method and image super-resolution equipment |
| CN106845449A (en) * | 2017-02-22 | 2017-06-13 | 浙江维尔科技有限公司 | A kind of image processing apparatus, method and face identification system |
| CN106951867A (en) * | 2017-03-22 | 2017-07-14 | 成都擎天树科技有限公司 | Face identification method, device, system and equipment based on convolutional neural networks |
Non-Patent Citations (1)
| Title |
|---|
| LI YI JUN等: "Generative Face Completion", 《COMPUTER VISION AND PATTERN RECOGNITION》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019015466A1 (en) * | 2017-07-17 | 2019-01-24 | 广州广电运通金融电子股份有限公司 | Method and apparatus for verifying person and certificate |
| US11763168B2 (en) | 2017-10-26 | 2023-09-19 | Nvidia Corporation | Progressive modification of generative adversarial neural networks |
| US11263525B2 (en) | 2017-10-26 | 2022-03-01 | Nvidia Corporation | Progressive modification of neural networks |
| US11250329B2 (en) | 2017-10-26 | 2022-02-15 | Nvidia Corporation | Progressive modification of generative adversarial neural networks |
| CN108090905A (en) * | 2018-01-05 | 2018-05-29 | 清华大学 | The determination methods and system of producing line exception |
| CN108280413A (en) * | 2018-01-17 | 2018-07-13 | 百度在线网络技术（北京）有限公司 | Face identification method and device |
| CN108280413B (en) * | 2018-01-17 | 2022-04-19 | 百度在线网络技术（北京）有限公司 | Face recognition method and device |
| CN108416326B (en) * | 2018-03-27 | 2021-07-16 | 百度在线网络技术（北京）有限公司 | Face recognition method and device |
| CN108416326A (en) * | 2018-03-27 | 2018-08-17 | 百度在线网络技术（北京）有限公司 | Face identification method and device |
| CN110263603A (en) * | 2018-05-14 | 2019-09-20 | 桂林远望智能通信科技有限公司 | Face identification method and device based on center loss and residual error visual simulation network |
| CN109711364A (en) * | 2018-12-29 | 2019-05-03 | 成都视观天下科技有限公司 | A kind of facial image super-resolution reconstruction method, device and computer equipment |
| CN111753595A (en) * | 2019-03-29 | 2020-10-09 | 北京市商汤科技开发有限公司 | Living body detection method and apparatus, device, and storage medium |
| WO2020199577A1 (en) * | 2019-03-29 | 2020-10-08 | 北京市商汤科技开发有限公司 | Method and device for living body detection, equipment, and storage medium |
| CN110163114B (en) * | 2019-04-25 | 2022-02-15 | 厦门瑞为信息技术有限公司 | Method and system for analyzing face angle and face blurriness and computer equipment |
| CN110163114A (en) * | 2019-04-25 | 2019-08-23 | 厦门瑞为信息技术有限公司 | A kind of facial angle and face method for analyzing ambiguity, system and computer equipment |
| CN110580682A (en) * | 2019-09-16 | 2019-12-17 | 电子科技大学 | Countermeasure network seismic data super-resolution reconstruction method based on optimization generation |
| CN111553208A (en) * | 2020-04-15 | 2020-08-18 | 上海携程国际旅行社有限公司 | Identity recognition method, system, device and medium based on image of people and certificate integration |
| CN112508782A (en) * | 2020-09-10 | 2021-03-16 | 浙江大华技术股份有限公司 | Network model training method, face image super-resolution reconstruction method and equipment |
| CN112508782B (en) * | 2020-09-10 | 2024-04-26 | 浙江大华技术股份有限公司 | Training method of network model, and super-resolution reconstruction method and device of face image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107423700B (en) | 2020-10-20 |
| WO2019015466A1 (en) | 2019-01-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107423700A (en) | The method and device of testimony verification | |
| Xu et al. | Locate globally, segment locally: A progressive architecture with knowledge review network for salient object detection | |
| CN107451607B (en) | A kind of personal identification method of the typical character based on deep learning | |
| CN105069746B (en) | Video real-time face replacement method and its system based on local affine invariant and color transfer technology | |
| CN107368787A (en) | A kind of Traffic Sign Recognition algorithm that application is driven towards depth intelligence | |
| CN112818862B (en) | Face tampering detection method and system based on multi-source clues and mixed attention | |
| CN108520503A (en) | A method of based on self-encoding encoder and generating confrontation network restoration face Incomplete image | |
| CN107609493A (en) | Optimize the method and device of face picture Environmental Evaluation Model | |
| CN106920243A (en) | The ceramic material part method for sequence image segmentation of improved full convolutional neural networks | |
| CN107392019A (en) | A kind of training of malicious code family and detection method and device | |
| CN106951840A (en) | A kind of facial feature points detection method | |
| CN107408211A (en) | Method for distinguishing is known again for object | |
| Cao et al. | Ancient mural restoration based on a modified generative adversarial network | |
| CN110490158A (en) | A kind of robust human face alignment schemes based on multistage model | |
| CN110689000B (en) | Vehicle license plate recognition method based on license plate sample generated in complex environment | |
| CN109063649A (en) | Pedestrian's recognition methods again of residual error network is aligned based on twin pedestrian | |
| CN109614856A (en) | Fungi image classification method based on convolutional neural networks | |
| CN113724354B (en) | Gray image coloring method based on reference picture color style | |
| CN109801225A (en) | Face reticulate pattern stain minimizing technology based on the full convolutional neural networks of multitask | |
| CN108961358A (en) | A kind of method, apparatus and electronic equipment obtaining samples pictures | |
| CN113537027A (en) | Face depth forgery detection method and system based on facial segmentation | |
| CN105956570A (en) | Lip characteristic and deep learning based smiling face recognition method | |
| CN103295019B (en) | A kind of Chinese fragment self-adaptive recovery method based on probability-statistics | |
| CN116805360B (en) | Obvious target detection method based on double-flow gating progressive optimization network | |
| CN109934796A (en) | A kind of automatic delineation method of organ based on Deep integrating study |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: No. 001-030, Yuntong space, office building, No. 9, Kelin Road, Science City, Guangzhou hi tech Industrial Development Zone, Guangzhou, Guangdong 510000 Patentee after: GRG TALLY-VISION I.T. Co.,Ltd. Country or region after: China Patentee after: Guangdian Yuntong Group Co.,Ltd. Address before: No. 001-030, Yuntong space, office building, No. 9, Kelin Road, Science City, Guangzhou hi tech Industrial Development Zone, Guangzhou, Guangdong 510000 Patentee before: GRG TALLY-VISION I.T. Co.,Ltd. Country or region before: China Patentee before: GRG BANKING EQUIPMENT Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240201 Address after: No. 001-030, Yuntong space, office building, No. 9, Kelin Road, Science City, Guangzhou hi tech Industrial Development Zone, Guangzhou, Guangdong 510000 Patentee after: GRG TALLY-VISION I.T. Co.,Ltd. Country or region after: China Address before: No. 001-030, Yuntong space, office building, No. 9, Kelin Road, Science City, Guangzhou hi tech Industrial Development Zone, Guangzhou, Guangdong 510000 Patentee before: GRG TALLY-VISION I.T. Co.,Ltd. Country or region before: China Patentee before: Guangdian Yuntong Group Co.,Ltd. |