CN110489659A - Data matching method and device - Google Patents
Data matching method and device Download PDFInfo
- Publication number
- CN110489659A CN110489659A CN201910651114.5A CN201910651114A CN110489659A CN 110489659 A CN110489659 A CN 110489659A CN 201910651114 A CN201910651114 A CN 201910651114A CN 110489659 A CN110489659 A CN 110489659A
- Authority
- CN
- China
- Prior art keywords
- vocal print
- print feature
- feature vector
- user
- target
- 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- 230000001755 vocal effect Effects 0.000 claims abstract description 287
- 239000013598 vector Substances 0.000 claims description 186
- 230000001815 facial effect Effects 0.000 claims description 90
- 238000009826 distribution Methods 0.000 claims description 44
- 238000000605 extraction Methods 0.000 claims description 25
- 238000004422 calculation algorithm Methods 0.000 claims description 23
- 238000012545 processing Methods 0.000 claims description 15
- 238000005286 illumination Methods 0.000 claims description 7
- 238000012216 screening Methods 0.000 claims description 4
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 239000000284 extract Substances 0.000 description 9
- 230000006870 function Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 230000003595 spectral effect Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000004590 computer program Methods 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/40—Information retrieval; Database structures therefor; File system structures therefor of multimedia data, e.g. slideshows comprising image and additional audio data
- G06F16/43—Querying
- G06F16/435—Filtering based on additional data, e.g. user or group profiles
- G06F16/436—Filtering based on additional data, e.g. user or group profiles using biological or physiological data of a human being, e.g. blood pressure, facial expression, gestures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Strategic Management (AREA)
- Physiology (AREA)
- Molecular Biology (AREA)
- Human Computer Interaction (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Multimedia (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention provides a kind of data matching method and devices.The present invention relates to artificial intelligence field, which includes: the request for receiving target user and issuing, and request carries voice data;The vocal print feature for extracting voice data, obtains target vocal print feature;The user stored in presetting database is screened according to target vocal print feature, obtain recommendation list, recommendation list includes an at least user, similarity between the vocal print feature and target vocal print feature of all users that recommendation list includes is greater than or equal to the first default similarity threshold, and the vocal print feature of multiple users is stored in presetting database;Recommendation list is shown to target user.Technical solution provided in an embodiment of the present invention is able to solve the problem of platform can not protect the individual privacy of user.
Description
[technical field]
The present invention relates to artificial intelligence field more particularly to a kind of data matching methods and device.
[background technique]
With popularizing for network, user increasingly gets used to network social intercourse, and user (can answer in social platform for social
With program) on input the age, geographical location, income, assets, the conditions such as educational background are screened, to find the people for oneself wanting to make friends.
But this method has a problem that: before adding good friend, the personal information of user can be checked by people on platform, platform
The individual privacy of user can not be protected.
[summary of the invention]
In view of this, can not be protected the embodiment of the invention provides a kind of data matching method and device to solve platform
The problem of protecting the individual privacy of user.
The embodiment of the invention provides a kind of data matching methods, which comprises receives asking for target user's sending
It asks, the request carries voice data;The vocal print feature for extracting the voice data obtains target vocal print feature;According to institute
It states target vocal print feature to screen the user stored in presetting database, obtains recommendation list, the recommendation list includes
An at least user, it is similar between the vocal print feature for all users that the recommendation list includes and the target vocal print feature
Degree is greater than or equal to the first default similarity threshold, and the vocal print feature of multiple users is stored in the presetting database;To institute
It states target user and shows the recommendation list.
Further, the vocal print feature for extracting the voice data, obtains target vocal print feature, comprising: from described
N kind vocal print feature vector is extracted in voice data, wherein N >=2;Any two kinds are calculated separately in the N kind vocal print feature vector
Average KL distance between vocal print feature vector;Average KL is special as the target vocal print apart from maximum two kinds of vocal print features
Sign.
Further, calculate the average KL distance between two kinds of vocal print feature vectors, comprising: obtain the first vocal print feature to
Amount and the second vocal print feature vector;Calculate separately the distribution of the second vocal print feature vector described in the first vocal print feature vector sum
Mean value and covariance;The mean value and covariance of the distribution of the second vocal print feature vector according to the first vocal print feature vector sum
Construct the first vocal print feature vector space and the corresponding probability distribution of the second vocal print characteristic vector space;According to
The first vocal print feature vector space and the corresponding probability distribution of the second vocal print characteristic vector space, described in calculating
Average KL distance between second vocal print feature vector described in first vocal print feature vector sum.
Further, the facial image of multiple users is also stored in the presetting database, the request also carries
First facial image, before the recommendation list to target user displaying, the method also includes: it calculates separately
Similarity between the facial image for each user that first facial image and the recommendation list include;By similarity
The corresponding user of facial image less than the second default similarity threshold deletes from the recommendation list.
Further, each user for calculating separately first facial image and the recommendation list includes
Similarity between facial image, comprising: illumination pretreatment, filter are carried out to first facial image using difference Gauss algorithm
Except the low-frequency information of first facial image, retains the high-frequency information of first facial image, obtain Gaussian image;To institute
It states Gaussian image and carries out image histogram equalization processing, obtain the uniform image of gray value;It is uniform to calculate the gray value
The corresponding feature vector of image, using the feature vector being calculated as the corresponding feature vector of first facial image;Point
The facial image for each user that the corresponding feature vector of first facial image and the recommendation list include is not calculated
Similarity between corresponding feature vector.
The embodiment of the invention provides a kind of data matching device, described device includes: receiving unit, for receiving target
The request that user issues, the request carry voice data;Extraction unit, the vocal print for extracting the voice data are special
Sign, obtains target vocal print feature;Screening unit, for according to the target vocal print feature to the user stored in presetting database
It is screened, obtains recommendation list, the recommendation list includes an at least user, all users that the recommendation list includes
Vocal print feature and the target vocal print feature between similarity be greater than or equal to the first default similarity threshold, it is described default
The vocal print feature of multiple users is stored in database;Display unit, for showing the recommendation list to the target user.
Further, the extraction unit includes: extraction subelement, for extracting N kind vocal print from the voice data
Feature vector, wherein N >=2;First computation subunit, for calculating separately any two kinds of sound in the N kind vocal print feature vector
Average KL distance between line feature vector;Subelement is determined, for the KL that will be averaged apart from maximum two kinds of vocal print feature conducts
The target vocal print feature.
Further, first computation subunit includes: acquisition module, for obtaining the first vocal print feature vector sum the
Two vocal print feature vectors;First computing module, for calculating separately the spy of the second vocal print described in the first vocal print feature vector sum
Levy the mean value and covariance of vector distribution;Module is constructed, the second vocal print according to the first vocal print feature vector sum is used for
The mean value and covariance of feature vector distribution construct the first vocal print feature vector space and the second vocal print feature vector
The corresponding probability distribution in space;Second computing module, for according to the first vocal print feature vector space and described the
It is special to calculate the second vocal print described in the first vocal print feature vector sum for the corresponding probability distribution of two vocal print feature vector spaces
Levy the average KL distance between vector.
Further, the facial image of multiple users is also stored in the presetting database, the request also carries
First facial image, described device further include: computing unit is used in the display unit to described in target user displaying
Before recommendation list, the facial image for each user that first facial image includes with the recommendation list is calculated separately
Between similarity;Unit is deleted, for the corresponding user of facial image by similarity less than the second default similarity threshold
It is deleted from the recommendation list.
Further, the computing unit includes: to filter out subelement, for utilizing difference Gauss algorithm to described the first
Face image carries out illumination pretreatment, filters out the low-frequency information of first facial image, retains the height of first facial image
Frequency information, obtains Gaussian image;Image histogram equalization processing subelement, for carrying out image histogram to the Gaussian image
Figure equalization processing obtains the uniform image of gray value;Second computation subunit, for calculating the uniform image of the gray value
Corresponding feature vector, using the feature vector being calculated as the corresponding feature vector of first facial image;Third meter
Operator unit, for calculate separately the corresponding feature vector of first facial image and the recommendation list includes each
Similarity between the corresponding feature vector of the facial image of user.
In embodiments of the present invention, the request that target user issues is received, request carries voice data, extracts voice number
According to vocal print feature, obtain target vocal print feature, the user stored in presetting database sieved according to target vocal print feature
Choosing has achieved the effect that carrying out commending friends to user according to vocal print feature can not check and use on platform before adding good friend
The personal information at family, the effective protection individual privacy of user.
[Detailed description of the invention]
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this field
For those of ordinary skill, without any creative labor, it can also be obtained according to these attached drawings other attached
Figure.
Fig. 1 is a kind of flow chart of optional data matching method according to embodiments of the present invention;
Fig. 2 is a kind of schematic diagram of optional data matching device according to embodiments of the present invention;
Fig. 3 is a kind of schematic diagram of computer equipment provided in an embodiment of the present invention.
[specific embodiment]
For a better understanding of the technical solution of the present invention, being retouched in detail to the embodiment of the present invention with reference to the accompanying drawing
It states.
It will be appreciated that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Base
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts it is all its
Its embodiment, shall fall within the protection scope of the present invention.
The term used in embodiments of the present invention is only to be not intended to be limiting merely for for the purpose of describing particular embodiments
The present invention.In the embodiment of the present invention and the "an" of singular used in the attached claims, " described " and "the"
It is also intended to including most forms, unless the context clearly indicates other meaning.
It should be appreciated that term "and/or" used herein is only a kind of incidence relation for describing affiliated partner, indicate
There may be three kinds of relationships, for example, A and/or B, can indicate: individualism A, exist simultaneously A and B, individualism B these three
Situation.In addition, character "/" herein, typicallys represent the relationship that forward-backward correlation object is a kind of "or".
The embodiment of the invention provides a kind of data matching method, this method can be applied to social platform.The party is used
The social platform of method can meet the needs of user is according to sound friend-making.
As an alternative embodiment, store the voice data of multiple users in social platform, work as target user
When wanting to make friends, target user issues to social platform and requests, and request carries voice data, and the vocal print for extracting voice data is special
Sign, obtains target vocal print feature, the stored and vocal print feature of target vocal print feature relatively is found out, then by these sound
The corresponding user of line feature recommends target user.
For example, user Xiao Ming wants to look for a sound in social platform as the people of some star makes friends, for the side of description
Just, which is known as AA, then Xiao Ming is target user, and Xiao Ming issues to social platform and requests, and request carries the voice of AA
Data (voice data can be the TV play from AA, film, interception obtains in variety), social platform extracts the voice of AA
The vocal print feature of data obtains target vocal print feature.Social platform has a database, which can be described as preset data
Library, the data of multiple users is stored in the database, and (including information such as age, geographical location, income, assets, educational background are also wrapped
Include the information such as vocal print feature and the facial image of user).Assuming that storing the data of 5000 users in the database, then from this
Screened in 5000 users several with target vocal print feature very close to user, these users are formed into recommendation list, and will
Recommendation list shows Xiao Ming.Xiao Ming can select one or more users from recommendation list, send addition good friend's request, other side
By the way that later, Xiao Ming becomes good friend with other side.Before adding good friend, Xiao Ming can not check the personal letter of user on platform
Breath, the effective protection individual privacy of user.
As another optional embodiment, used the social activity of data matching method provided in an embodiment of the present invention flat
Platform can meet the needs of user is according to audio+photo friend-making.Store the voice data and face of multiple users in social platform
Image, when target user wants to make friends, target user issues to social platform and requests, and request carries voice data and face
The facial image that request carries (for convenience, is known as the first facial image) by image, and the vocal print for extracting voice data is special
Sign, obtains target vocal print feature, the stored and vocal print feature of target vocal print feature relatively is found out, then from these sound
Facial image and the close user of the first facial image are found out in the corresponding user of line feature, and recommend target user.
For example, user Xiao Ming wants to look for a sound as AA star in social platform and face looks like the people of BB star
It makes friends, then Xiao Ming is target user, and Xiao Ming issues to social platform and requests, and request carries voice data (voice of AA
Data can be that the TV play from AA, film, interception obtains in variety) and BB facial image (i.e. above-mentioned first face figure
Picture), social platform extracts the vocal print feature of the voice data of AA, obtains target vocal print feature.Social platform has a data
Library, the database can be described as presetting database, stored in the database multiple users data (including the age, geographical location,
The information such as income, assets, educational background also include the information such as vocal print feature and the facial image of user).Assuming that being stored in the database
The data of 5000 users, then screened from this 5000 users several with target vocal print feature very close to user, will
These users form recommendation list, it is assumed that have 50 users in recommendation list, then find out face from this 50 users and look like BB
User, it is assumed that find out is No. 0009 user, No. 0078 user, No. 4560 users, then by No. 0009 user, No. 0078 with
Family, No. 4560 users show Xiao Ming.Xiao Ming can select one or more users from recommendation list, send addition good friend and ask
It asks, after other side passes through, Xiao Ming becomes good friend with other side.Before adding good friend, Xiao Ming can not check user on platform
Personal information, the effective protection individual privacy of user.Assuming that Xiao Ming has sent addition good friend's request to No. 0078 user, then
After No. 0078 user receives the request, the data of Xiao Ming can be checked, to decide whether Xiao Ming being added to good friend.If No. 0078
User agree to addition good friend, then after, Xiao Ming can check the data of No. 0078 user.
In embodiments of the present invention, the voice data and facial image of the multiple users stored in social platform be user from
What own wish uploaded, there is no the voice data and facial image in the unwitting situation lower platform acquisition user of user to cause to invade
The problem of violating privacy of user.
Fig. 1 is a kind of flow chart of optional data matching method according to embodiments of the present invention.As shown in Figure 1, this method
Include:
Step S102, receives the request that target user issues, and request carries voice data.
Step S104 extracts the vocal print feature of voice data, obtains target vocal print feature.
Step S106 screens the user stored in presetting database according to target vocal print feature, obtains recommending name
Single, recommendation list includes an at least user, the vocal print feature and target vocal print feature of all users that recommendation list includes it
Between similarity be greater than or equal to the first default similarity threshold, the vocal print feature of multiple users is stored in presetting database.
Step S108 shows recommendation list to target user.
In embodiments of the present invention, the request that target user issues is received, request carries voice data, extracts voice number
According to vocal print feature, obtain target vocal print feature, the user stored in presetting database sieved according to target vocal print feature
Choosing has achieved the effect that carrying out commending friends to user according to vocal print feature can not check and use on platform before adding good friend
The personal information at family, the effective protection individual privacy of user.
Inventors have found that original life cannot be represented completely using the feature vector that single vocal print feature extraction algorithm extracts
The characteristics of object sample, therefore in embodiments of the present invention, inventors herein propose using at least two vocal print feature extraction algorithms come
The method for representing primitive organism sample.Specifically, after receiving the request that target user issues, from the voice data of request carrying
Middle extraction N kind vocal print feature vector, wherein N >=2;Calculate separately any two kinds of vocal print feature vectors in N kind vocal print feature vector
Between average KL distance;By average KL apart from maximum two kinds of vocal print features as target vocal print feature.
Wherein, the average KL distance between two kinds of vocal print feature vectors is calculated, detailed process can be with are as follows: obtains the first vocal print
Feature vector and the second vocal print feature vector;Calculate separately the equal of first vocal print feature vector sum the second vocal print feature vector distribution
Value and covariance;The first sound is constructed according to the mean value of first vocal print feature vector sum the second vocal print feature vector distribution and covariance
Line characteristic vector space and the corresponding probability distribution of the second vocal print characteristic vector space;It is empty according to the first vocal print feature vector
Between probability distribution corresponding with the second vocal print characteristic vector space, calculate first the second vocal print feature of vocal print feature vector sum
Average KL distance between vector.
Different types of vocal print feature vector can be extracted using different vocal print feature extraction algorithms, for example, can adopt
Vocal print feature extraction algorithm has: MFCC mel cepstrum coefficients, the phase angle residual phase residual error, LPCC linear prediction
Spectral function, the linear spectral function of MLSF Meier.
Specifically, the mean μ of the first vocal print feature vector distribution is calculated according to formula (1)A;The is calculated according to formula (2)
The mean μ of two vocal print feature vector distributionsB;The covariance cov of the first vocal print feature vector distribution is calculated according to formula (3)A;Root
The covariance cov of the second vocal print feature vector distribution is calculated according to formula (4)B;The first vocal print feature vector is calculated according to formula (5)
The corresponding probability distribution P in spaceA(x);The corresponding probability distribution P of the second vocal print characteristic vector space is calculated according to formula (6)B
(x);The average KL distance D between first vocal print feature vector sum the second vocal print feature vector is calculated according to formula (7).
Formula (1) is into (7), μAIndicate the mean value of the first vocal print feature vector distribution;μBIndicate the first vocal print feature vector
The mean value of distribution;covAIndicate the covariance of the first vocal print feature vector distribution;covBIndicate the distribution of the second vocal print feature vector
Covariance;PA(x) the corresponding probability distribution of the first vocal print feature vector space is indicated;PB(x) indicate that the second vocal print feature vector is empty
Between corresponding probability distribution;D indicates the average KL distance between first vocal print feature vector sum the second vocal print feature vector;nATable
Show the quantity for the feature vector that the first vocal print feature vector includes, wherein the ith feature that the first vocal print feature vector includes to
Amount is xAi, i takes 1 to nABetween natural number;nBIndicate the quantity for the feature vector that the second vocal print feature vector includes, wherein
The ith feature vector that second vocal print feature vector includes is xBi, i takes 1 to nBBetween natural number.
A variety of vocal print features of each user are stored in presetting database, wherein each vocal print feature and one kind
Algorithm is corresponding.For example, storing the vocal print feature of m user in presetting database, this m user is respectively user U1, user
U2, user U3 ..., user Um store N kind in presetting database also, for each user in this m user
Vocal print feature, wherein the 1st kind of vocal print feature of user Uj is extracted according to vocal print feature extraction algorithm S1, user Uj
The 2nd kind of vocal print feature be to be extracted according to vocal print feature extraction algorithm S2, the 3rd kind of vocal print feature of user Uj is basis
Vocal print feature extraction algorithm S3 is extracted ... ..., and the N kind vocal print feature of user Uj is according to vocal print feature extraction algorithm
What SN was extracted, j takes 1 to the random natural number between m.Vocal print feature extraction algorithm S1, vocal print feature extraction algorithm
S2 ..., vocal print feature extraction algorithm SN refer specifically to MFCC mel cepstrum coefficients described above, residual phase
Phase angle residual error, LPCC linear prediction spectral function, the linear spectral function of MLSF Meier etc..
It is calculated separately in N kind vocal print feature vector between any two kinds of vocal print feature vectors according to above-mentioned formula (1) to (7)
Average KL distance;By average KL apart from maximum two kinds of vocal print features as target vocal print feature.Assuming that average KL distance is most
Two kinds of big vocal print features are vocal print feature C1 and vocal print feature C2, and vocal print feature C1 is according to vocal print feature extraction algorithm
What S1 was extracted, vocal print feature C2 is extracted according to vocal print feature extraction algorithm S2, then by vocal print feature C1 harmony
Line feature C2 is used as target vocal print feature.
In step S106, during being screened according to target vocal print feature to the user stored in presetting database,
The similarity for first calculating the vocal print feature of the user stored in target vocal print feature and presetting database, is then sieved according to similarity
Family is selected, for example, filtering out the user that similarity is greater than or equal to the first default similarity threshold.
Illustrate the process for calculating similarity by taking user's first as an example.
Calculate the tool of the similarity simi of the vocal print feature of the user's first stored in target vocal print feature and presetting database
Body process is as follows: calculating similarity simi according to formula simi=(simi1+simi2)/2, wherein simi1 is vocal print feature C1
Similarity between the vocal print feature for the user's first extracted according to algorithm S1, simi2 are vocal print feature C2 and according to calculations
Similarity between the vocal print feature for user's first that method S2 is extracted.
The average KL of two kinds of vocal print feature vectors apart from it is larger when, the correlation of both vocal print feature vectors is smaller.In
When carrying out the similarity calculation of vocal print feature, it is special to calculate separately the vocal print stored in two kinds of vocal print feature vectors and presetting database
The similarity between vector is levied, can be avoided single vocal print feature extraction algorithm bring error when extracting vocal print feature, from
And effectively improve the accuracy for calculating similarity.
In embodiments of the present invention, the user of social platform can will be used with upload pictures (i.e. facial image), social platform
The photo that family uploads is stored into presetting database.When upload pictures, user provides ID card information, and platform, which carries out identity, to be recognized
Card terminates if certification does not pass through;If certification passes through, the photo of user's upload is received.The photograph that platform uploads user
Piece and the identity card picture of user compare, and judge whether to match, if it does, then the photo that user is uploaded saves;If no
Matching then asks user to upload my photo or check whether photo passes through PS.To sum up, user is without serious PS's
Photo can be just successfully saved in the photo library of the user;The photo of photo or serious distortion if not user is all
It will not be saved in the photo library of the user, the photo in photo library to guarantee user is strictly the photo of user,
And these photos can compare the appearance for being truly reflected user.
As an alternative embodiment, the request that target user issues not only carries voice data, also carry
First facial image, this illustrates that target user wishes to find good friend, in this case, platform by way of voice+photo
Recommendation list is being filtered out according to voice and then the user in recommendation list is further being screened according to photo, specifically
Ground, the similarity for calculating separately the first facial image between the facial image of each user that recommendation list includes;By phase
The corresponding user of facial image like degree less than the second default similarity threshold deletes from recommendation list.
Inventors have found that during searching similar face image, if two facial images compared are all positive dough figurines
The similarity-rough set of face, then two facial images calculated is accurate;If two facial images compared have one or two to be
Side face, then the similarity-rough set inaccuracy of two facial images calculated.
The face figure used in order to improve the accuracy of comparison, when as an alternative embodiment, calculating similarity
As being front face image.Strong classifier or identification model can be used to identify facial image whether for front face image.
If the first facial image that strong classifier or identification model identify that target user uploads is not front face image, remind
Target user uploads front face image, until obtaining front face image.
Establish the specific steps of strong classifier are as follows: extract front face image as training positive sample, extract side face
Image extracts integrating channel feature as training negative sample, trains strong classification from extraction feature using Adaboost algorithm
Device.Strong classifier for giving a mark to face, score it is low be side face image, score it is high be front face image.
Establish the specific steps of identification model are as follows: face samples pictures are pre-processed to obtain the gray level image of unified pixel,
Then the gray level image of unified pixel is divided into face image and side face image;Using obtained face image as unsupervised feature
The input for learning PCANet carries out unsupervised positive face feature learning, obtains positive face feature;Using obtained side face image as there is prison
Educational inspector practises the input of CNN, and combines obtained positive face feature, establishes side face feature and just by the processing of supervised learning CNN
Mapping relations between face feature;Unified positive face feature is obtained using mapping relations, unified positive face feature is sent into supporting vector
The training that identification model is carried out in machine, obtains identification model.
Optionally, the first facial image is calculated separately between the facial image of each user that recommendation list includes
Similarity, comprising: illumination pretreatment is carried out to the first facial image using difference Gauss algorithm, filters out the low of the first facial image
Frequency information retains the high-frequency information of the first facial image, obtains Gaussian image;Image histogram equalization is carried out to Gaussian image
Processing, obtains the uniform image of gray value;Calculate the corresponding feature vector of the uniform image of gray value, the feature that will be calculated
Vector is as the corresponding feature vector of the first facial image;It calculates separately the corresponding feature vector of the first facial image and recommends name
Similarity between the corresponding feature vector of the facial image for each user for singly including.
Image histogram equalization processing is the grey level histogram of image from the shape for comparing concentration in some gray scale interval
Formula becomes equally distributed form in whole tonal ranges, to increase the local contrast of image, makes the part of image
It is more clear.By handling the first facial image before calculating similarity, two facial images of calculating are improved
Between similarity accuracy.
The embodiment of the invention also provides a kind of data matching device, the device is for executing above-mentioned data matching method.
As shown in Fig. 2, the device includes: receiving unit 12, extraction unit 14, screening unit 16, display unit 18.
Receiving unit 12, for receiving the request of target user's sending, request carries voice data.
Extraction unit 14 obtains target vocal print feature for extracting the vocal print feature of voice data.
Screening unit 16 is obtained for being screened according to target vocal print feature to the user stored in presetting database
Recommendation list, recommendation list include an at least user, the vocal print feature and target vocal print of all users that recommendation list includes
Similarity between feature is greater than or equal to the first default similarity threshold, and the vocal print of multiple users is stored in presetting database
Feature.
Display unit 18, for showing recommendation list to target user.
Optionally, extraction unit 14 includes: to extract subelement, the first computation subunit, determine subelement.It is single to extract son
Member, for extracting N kind vocal print feature vector from voice data, wherein N >=2.First computation subunit, for calculating separately N
Average KL distance in kind vocal print feature vector between any two kinds of vocal print feature vectors.Subelement is determined, for the KL that will be averaged
Apart from maximum two kinds of vocal print features as target vocal print feature.
Optionally, the first computation subunit includes: to obtain module, the first computing module, building module, the second calculating mould
Block.Module is obtained, for obtaining first vocal print feature vector sum the second vocal print feature vector.First computing module, for distinguishing
Calculate the mean value and covariance of the distribution of first vocal print feature vector sum the second vocal print feature vector.Module is constructed, for according to the
One vocal print feature vector sum the second vocal print feature vector distribution mean value and covariance construct the first vocal print feature vector space and
The corresponding probability distribution of second vocal print characteristic vector space.Second computing module, for according to the first vocal print feature vector
It is special to calculate first the second vocal print of vocal print feature vector sum for space and the corresponding probability distribution of the second vocal print characteristic vector space
Levy the average KL distance between vector.
Optionally, the facial image of multiple users is also stored in presetting database, request also carries the first face figure
Picture, device further include: computing unit deletes unit.Computing unit recommends name for showing in display unit 18 to target user
Before list, the similarity that calculates separately the first facial image between the facial image of each user that recommendation list includes.
Unit is deleted, is deleted from recommendation list for the corresponding user of facial image by similarity less than the second default similarity threshold
It removes.
Optionally, computing unit includes: to filter out subelement, image histogram equalization processing subelement, the second calculating
Unit, third computation subunit.Subelement is filtered out, is located in advance for carrying out illumination to the first facial image using difference Gauss algorithm
Reason filters out the low-frequency information of the first facial image, retains the high-frequency information of the first facial image, obtains Gaussian image.Image is straight
It is uniform to obtain gray value for carrying out image histogram equalization processing to Gaussian image for square figure equalization processing subelement
Image.Second computation subunit, for calculating the corresponding feature vector of the uniform image of gray value, by the feature being calculated to
Amount is used as the corresponding feature vector of the first facial image.Third computation subunit, it is corresponding for calculating separately the first facial image
Feature vector feature vector corresponding with the facial image for each user that recommendation list includes between similarity.
The embodiment of the invention provides a kind of storage medium, storage medium includes the program of storage, wherein is run in program
When control storage medium where equipment execute following steps: receive target user issue request, request carry voice data;
The vocal print feature for extracting voice data, obtains target vocal print feature;According to target vocal print feature to storing in presetting database
User screens, and obtains recommendation list, and recommendation list includes an at least user, the sound for all users that recommendation list includes
Similarity between line feature and target vocal print feature is greater than or equal to the first default similarity threshold, stores in presetting database
The vocal print feature of multiple users;Recommendation list is shown to target user.
Optionally, when program is run, equipment where control storage medium also executes following steps: mentioning from voice data
Take N kind vocal print feature vector, wherein N >=2;It calculates separately in N kind vocal print feature vector between any two kinds of vocal print feature vectors
Average KL distance;By average KL apart from maximum two kinds of vocal print features as target vocal print feature.
Optionally, when program is run, equipment where control storage medium also executes following steps: it is special to obtain the first vocal print
Levy the second vocal print of vector sum feature vector;Calculate separately the mean value of first vocal print feature vector sum the second vocal print feature vector distribution
With covariance;The first vocal print is constructed according to the mean value of first vocal print feature vector sum the second vocal print feature vector distribution and covariance
Characteristic vector space and the corresponding probability distribution of the second vocal print characteristic vector space;According to the first vocal print feature vector space
Probability distribution corresponding with the second vocal print characteristic vector space, calculate first the second vocal print feature of vocal print feature vector sum to
Average KL distance between amount.
Optionally, when program is run, equipment where control storage medium also executes following steps: to target user's exhibition
Before showing recommendation list, the first facial image is calculated separately between the facial image of each user that recommendation list includes
Similarity;The corresponding user of facial image by similarity less than the second default similarity threshold deletes from recommendation list.
Optionally, when program is run, equipment where control storage medium also executes following steps: being calculated using difference Gauss
Method carries out illumination pretreatment to the first facial image, filters out the low-frequency information of the first facial image, retains the first facial image
High-frequency information obtains Gaussian image;Image histogram equalization processing is carried out to Gaussian image, gray value is obtained and uniformly schemes
Picture;The corresponding feature vector of the uniform image of gray value is calculated, using the feature vector being calculated as the first facial image pair
The feature vector answered;Calculate separately the people for each user that the corresponding feature vector of the first facial image and recommendation list include
Similarity between the corresponding feature vector of face image.
The embodiment of the invention provides a kind of computer equipments, including memory and processor, and memory is for storing packet
The information of program instruction is included, processor is used to control the execution of program instruction, real when program instruction is loaded and executed by processor
Existing following steps: receiving the request that target user issues, and request carries voice data;The vocal print feature of voice data is extracted,
Obtain target vocal print feature;The user stored in presetting database is screened according to target vocal print feature, obtains recommending name
Single, recommendation list includes an at least user, the vocal print feature and target vocal print feature of all users that recommendation list includes it
Between similarity be greater than or equal to the first default similarity threshold, the vocal print feature of multiple users is stored in presetting database;
Recommendation list is shown to target user.
Optionally, the extraction N kind from voice data is also performed the steps of when program instruction is loaded and executed by processor
Vocal print feature vector, wherein N >=2;It calculates separately flat between any two kinds of vocal print feature vectors in N kind vocal print feature vector
Equal KL distance;By average KL apart from maximum two kinds of vocal print features as target vocal print feature.
Optionally, when program instruction is loaded and is executed by processor also perform the steps of obtain the first vocal print feature to
Amount and the second vocal print feature vector;Calculate separately mean value and the association of the distribution of first vocal print feature vector sum the second vocal print feature vector
Variance;The first vocal print feature is constructed according to the mean value of first vocal print feature vector sum the second vocal print feature vector distribution and covariance
Vector space and the corresponding probability distribution of the second vocal print characteristic vector space;According to the first vocal print feature vector space and
The corresponding probability distribution of two vocal print feature vector spaces, calculate first the second vocal print of vocal print feature vector sum feature vector it
Between average KL distance.
Optionally, it also performs the steps of when program instruction is loaded and executed by processor and is pushed away to target user's displaying
Before recommending list, calculate separately similar between the first facial image and the facial image for each user that recommendation list includes
Degree;The corresponding user of facial image by similarity less than the second default similarity threshold deletes from recommendation list.
Optionally, it is also performed the steps of when program instruction is loaded and executed by processor and utilizes difference Gauss algorithm pair
First facial image carries out illumination pretreatment, filters out the low-frequency information of the first facial image, retains the high frequency of the first facial image
Information obtains Gaussian image;Image histogram equalization processing is carried out to Gaussian image, obtains the uniform image of gray value;Meter
The corresponding feature vector of the uniform image of gray value is calculated, using the feature vector being calculated as the corresponding spy of the first facial image
Levy vector;Calculate separately the facial image for each user that the corresponding feature vector of the first facial image and recommendation list include
Similarity between corresponding feature vector.
Fig. 3 is a kind of schematic diagram of computer equipment provided in an embodiment of the present invention.As shown in figure 3, the meter of the embodiment
Machine equipment 50 is calculated to include: processor 51, memory 52 and be stored in the meter that can be run in memory 52 and on processor 51
Calculation machine program 53 realizes the data matching method in embodiment when the computer program 53 is executed by processor 51, to avoid weight
It is multiple, it does not repeat one by one herein.Alternatively, being realized when the computer program is executed by processor 51 in embodiment in data matching device
The function of each model/unit does not repeat one by one herein to avoid repeating.
Computer equipment 50 can be desktop PC, notebook, palm PC and cloud server etc. and calculate equipment.
Computer equipment may include, but be not limited only to, processor 51, memory 52.It will be understood by those skilled in the art that Fig. 3 is only
It is the example of computer equipment 50, does not constitute the restriction to computer equipment 50, may include more more or fewer than illustrating
Component perhaps combines certain components or different components, such as computer equipment can also include input-output equipment, net
Network access device, bus etc..
Alleged processor 51 can be central processing unit (Central Processing Unit, CPU), can also be
Other general processors, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit
(Application Specific Integrated Circuit, ASIC), field programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processor
Deng.
Memory 52 can be the internal storage unit of computer equipment 50, such as the hard disk or interior of computer equipment 50
It deposits.Memory 52 is also possible to the plug-in type being equipped on the External memory equipment of computer equipment 50, such as computer equipment 50
Hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash card
(Flash Card) etc..Further, memory 52 can also both including computer equipment 50 internal storage unit and also including
External memory equipment.Memory 52 is for storing other programs and data needed for computer program and computer equipment.It deposits
Reservoir 52 can be also used for temporarily storing the data that has exported or will export.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In several embodiments provided by the present invention, it should be understood that disclosed system, device and method can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit
It divides, only a kind of logical function partition, there may be another division manner in actual implementation, for example, multiple units or group
Part can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown
Or the mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, device or unit it is indirect
Coupling or communication connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.
The above-mentioned integrated unit being realized in the form of SFU software functional unit can store and computer-readable deposit at one
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
It is each that device (can be personal computer, server or network equipment etc.) or processor (Processor) execute the present invention
The part steps of embodiment the method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (Read-
Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic or disk etc. it is various
It can store the medium of program code.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the present invention.
Claims (10)
1. a kind of data matching method, which is characterized in that the described method includes:
The request that target user issues is received, the request carries voice data;
The vocal print feature for extracting the voice data obtains target vocal print feature;
The user stored in presetting database is screened according to the target vocal print feature, obtains recommendation list, it is described to push away
Recommending list includes an at least user, the vocal print feature for all users that the recommendation list includes and the target vocal print feature
Between similarity be greater than or equal to the first default similarity threshold, the vocal print of multiple users is stored in the presetting database
Feature;
The recommendation list is shown to the target user.
2. the method according to claim 1, wherein the vocal print feature for extracting the voice data, obtains
Target vocal print feature, comprising:
N kind vocal print feature vector is extracted from the voice data, wherein N >=2;
Calculate separately the average KL distance in the N kind vocal print feature vector between any two kinds of vocal print feature vectors;
By average KL apart from maximum two kinds of vocal print features as the target vocal print feature.
3. according to the method described in claim 2, it is characterized in that, calculate two kinds of vocal print feature vectors between average KL away from
From, comprising:
Obtain first vocal print feature vector sum the second vocal print feature vector;
Calculate separately the mean value and covariance of the distribution of the second vocal print feature vector described in the first vocal print feature vector sum;
Described in the mean value and covariance building of the distribution of the second vocal print feature vector according to the first vocal print feature vector sum
First vocal print feature vector space and the corresponding probability distribution of the second vocal print characteristic vector space;
According to the first vocal print feature vector space and the corresponding probability distribution of the second vocal print characteristic vector space,
Calculate the average KL distance between the second vocal print feature vector described in the first vocal print feature vector sum.
4. the method according to claim 1, wherein the people of multiple users is also stored in the presetting database
Face image, the request also carry the first facial image, before the recommendation list to target user displaying,
The method also includes:
Calculate separately the phase between the facial image for each user that first facial image includes with the recommendation list
Like degree;
The corresponding user of facial image by similarity less than the second default similarity threshold deletes from the recommendation list.
5. according to the method described in claim 4, it is characterized in that, described calculate separately first facial image and push away with described
Recommend the similarity between the facial image for each user that list includes, comprising:
Illumination pretreatment is carried out to first facial image using difference Gauss algorithm, filters out the low of first facial image
Frequency information retains the high-frequency information of first facial image, obtains Gaussian image;
Image histogram equalization processing is carried out to the Gaussian image, obtains the uniform image of gray value;
The corresponding feature vector of the uniform image of the gray value is calculated, using the feature vector being calculated as described the first
The corresponding feature vector of face image;
Calculate separately the people for each user that the corresponding feature vector of first facial image and the recommendation list include
Similarity between the corresponding feature vector of face image.
6. a kind of data matching device, which is characterized in that described device includes:
Receiving unit, for receiving the request of target user's sending, the request carries voice data;
Extraction unit obtains target vocal print feature for extracting the vocal print feature of the voice data;
Screening unit is pushed away for being screened according to the target vocal print feature to the user stored in presetting database
List is recommended, the recommendation list includes an at least user, the vocal print feature for all users that the recommendation list includes and institute
It states the similarity between target vocal print feature and is greater than or equal to the first default similarity threshold, stored in the presetting database
The vocal print feature of multiple users;
Display unit, for showing the recommendation list to the target user.
7. device according to claim 6, which is characterized in that the extraction unit includes:
Subelement is extracted, for extracting N kind vocal print feature vector from the voice data, wherein N >=2;
First computation subunit, for calculating separately in the N kind vocal print feature vector between any two kinds of vocal print feature vectors
Average KL distance;
Subelement is determined, for the KL that will be averaged apart from maximum two kinds of vocal print features as the target vocal print feature.
8. device according to claim 7, which is characterized in that first computation subunit includes:
Module is obtained, for obtaining first vocal print feature vector sum the second vocal print feature vector;
First computing module, for calculating separately the distribution of the second vocal print feature vector described in the first vocal print feature vector sum
Mean value and covariance;
Module is constructed, mean value and association for the distribution of the second vocal print feature vector according to the first vocal print feature vector sum
Variance constructs the first vocal print feature vector space and the corresponding probability distribution of the second vocal print characteristic vector space;
Second computing module, for according to the first vocal print feature vector space and the second vocal print characteristic vector space point
Not corresponding probability distribution, calculate average KL between the second vocal print feature vector described in the first vocal print feature vector sum away from
From.
9. a kind of storage medium, which is characterized in that the storage medium includes the program of storage, wherein run in described program
When control the storage medium where equipment perform claim require any one of 1 to 5 described in data matching method.
10. a kind of computer equipment, including memory and processor, the memory is for storing the letter including program instruction
Breath, the processor are used to control the execution of program instruction, it is characterised in that: described program instruction is loaded and executed by processor
The step of data matching method described in Shi Shixian claim 1 to 5 any one.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910651114.5A CN110489659A (en) | 2019-07-18 | 2019-07-18 | Data matching method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910651114.5A CN110489659A (en) | 2019-07-18 | 2019-07-18 | Data matching method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110489659A true CN110489659A (en) | 2019-11-22 |
Family
ID=68547404
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910651114.5A Pending CN110489659A (en) | 2019-07-18 | 2019-07-18 | Data matching method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110489659A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111816191A (en) * | 2020-07-08 | 2020-10-23 | 珠海格力电器股份有限公司 | Voice processing method, device, system and storage medium |
CN112270927A (en) * | 2020-09-27 | 2021-01-26 | 青岛海尔空调器有限总公司 | Intelligent interaction method based on environment adjusting equipment and intelligent interaction equipment |
CN112818095A (en) * | 2021-01-26 | 2021-05-18 | 广州欢网科技有限责任公司 | Method, device and equipment for matching and associating questions |
CN113239229A (en) * | 2021-06-17 | 2021-08-10 | 张鹏涛 | Intelligent screening data processing method and system and cloud platform |
CN113299296A (en) * | 2021-05-08 | 2021-08-24 | 深圳市沃特沃德信息有限公司 | Friend making method and device based on voiceprint recognition and computer equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20100095166A (en) * | 2009-02-20 | 2010-08-30 | 성균관대학교산학협력단 | A personal adaptive music recommendation method using analysis of playlists of users |
CN104183245A (en) * | 2014-09-04 | 2014-12-03 | 福建星网视易信息系统有限公司 | Method and device for recommending music stars with tones similar to those of singers |
CN105575393A (en) * | 2015-12-02 | 2016-05-11 | 中国传媒大学 | Personalized song recommendation method based on voice timbre |
CN105656756A (en) * | 2015-12-28 | 2016-06-08 | 百度在线网络技术(北京)有限公司 | Friend recommendation method and device |
CN108495143A (en) * | 2018-03-30 | 2018-09-04 | 百度在线网络技术(北京)有限公司 | The method and apparatus of video recommendations |
-
2019
- 2019-07-18 CN CN201910651114.5A patent/CN110489659A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20100095166A (en) * | 2009-02-20 | 2010-08-30 | 성균관대학교산학협력단 | A personal adaptive music recommendation method using analysis of playlists of users |
CN104183245A (en) * | 2014-09-04 | 2014-12-03 | 福建星网视易信息系统有限公司 | Method and device for recommending music stars with tones similar to those of singers |
CN105575393A (en) * | 2015-12-02 | 2016-05-11 | 中国传媒大学 | Personalized song recommendation method based on voice timbre |
CN105656756A (en) * | 2015-12-28 | 2016-06-08 | 百度在线网络技术(北京)有限公司 | Friend recommendation method and device |
CN108495143A (en) * | 2018-03-30 | 2018-09-04 | 百度在线网络技术(北京)有限公司 | The method and apparatus of video recommendations |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111816191A (en) * | 2020-07-08 | 2020-10-23 | 珠海格力电器股份有限公司 | Voice processing method, device, system and storage medium |
WO2022007497A1 (en) * | 2020-07-08 | 2022-01-13 | 珠海格力电器股份有限公司 | Voice processing method and apparatus, system and storage medium |
CN112270927A (en) * | 2020-09-27 | 2021-01-26 | 青岛海尔空调器有限总公司 | Intelligent interaction method based on environment adjusting equipment and intelligent interaction equipment |
CN112818095A (en) * | 2021-01-26 | 2021-05-18 | 广州欢网科技有限责任公司 | Method, device and equipment for matching and associating questions |
CN113299296A (en) * | 2021-05-08 | 2021-08-24 | 深圳市沃特沃德信息有限公司 | Friend making method and device based on voiceprint recognition and computer equipment |
CN113239229A (en) * | 2021-06-17 | 2021-08-10 | 张鹏涛 | Intelligent screening data processing method and system and cloud platform |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108197532B (en) | The method, apparatus and computer installation of recognition of face | |
CN110489659A (en) | Data matching method and device | |
CN109284733B (en) | Shopping guide negative behavior monitoring method based on yolo and multitask convolutional neural network | |
US10810870B2 (en) | Method of processing passage record and device | |
WO2020258667A1 (en) | Image recognition method and apparatus, and non-volatile readable storage medium and computer device | |
CN108197250B (en) | Picture retrieval method, electronic equipment and storage medium | |
WO2021082118A1 (en) | Person re-identification method and apparatus, and terminal and storage medium | |
CN111738357B (en) | Junk picture identification method, device and equipment | |
CN108269254A (en) | Image quality measure method and apparatus | |
CN110222728B (en) | Training method and system of article identification model and article identification method and equipment | |
CN110428399A (en) | Method, apparatus, equipment and storage medium for detection image | |
CN110163111A (en) | Method, apparatus of calling out the numbers, electronic equipment and storage medium based on recognition of face | |
CN112163637B (en) | Image classification model training method and device based on unbalanced data | |
CN109871845A (en) | Certificate image extracting method and terminal device | |
CN109902561A (en) | A kind of face identification method and device, robot applied to robot | |
CN110689046A (en) | Image recognition method, image recognition device, computer device, and storage medium | |
CN113515988A (en) | Palm print recognition method, feature extraction model training method, device and medium | |
CN110245573A (en) | A kind of register method, apparatus and terminal device based on recognition of face | |
CN110287767A (en) | Can attack protection biopsy method, device, computer equipment and storage medium | |
CN112329586A (en) | Client return visit method and device based on emotion recognition and computer equipment | |
CN111784665A (en) | OCT image quality assessment method, system and device based on Fourier transform | |
CN108090108A (en) | Information processing method, device, electronic equipment and storage medium | |
CN113157956B (en) | Picture searching method, system, mobile terminal and storage medium | |
CN112651333A (en) | Silence living body detection method and device, terminal equipment and storage medium | |
CN112614110A (en) | Method and device for evaluating image quality and terminal equipment |
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 |