CN107808149A

CN107808149A - A kind of face information mask method, device and storage medium

Info

Publication number: CN107808149A
Application number: CN201711148022.2A
Authority: CN
Inventors: 张亿皓; 李绍欣; 陈志博; 吴永坚; 黄飞跃
Original assignee: Tencent Cyber Tianjin Co Ltd
Current assignee: Tencent Cyber Tianjin Co Ltd
Priority date: 2017-11-17
Filing date: 2017-11-17
Publication date: 2018-03-16

Abstract

The embodiment of the invention discloses a kind of face information mask method, device and storage medium；Face information to be identified is identified according to default verification sample set for the embodiment of the present invention, after being identified result, can be according to face information to be identified indicated in recognition result and the matching degree of destination object (verify in sample set and verify sample with the face information matching degree highest to be identified), to be labeled to face information to be identified, and the destination object in verification sample set is updated according to face information after mark, face can be subsequently identified using the destination object after the renewal as foundation；The program can not only improve the treatment effeciency of mark, furthermore, it is possible to avoid maloperation, improve recognition effect.

Description

A kind of face information mask method, device and storage medium

Technical field

The present invention relates to communication technical field, and in particular to a kind of face information mask method, device and storage medium.

Background technology

Recognition of face, also referred to as face recognition or face recognition, it is that the facial feature information based on people carries out identification A kind of biological identification technology.It specifically can gather image or video flowing containing face by picture pick-up device, and exist automatically Detect and track face in image, and then the face detected is matched with default verification sample, to carry out identity knowledge Not.Because face can change according to the passage of time, therefore, it is necessary to verification sample is updated in time, Cai Nengbao The accuracy of recognition of face is demonstrate,proved, to prevent identifying system from obvious performance degradation phenomenon occur.

In the prior art, typically all can by carrying out manual examination and verification and mark to the face information that collects, and according to Mark selects suitable face information, to be updated to verification sample, for example, specifically can be by manually being gathered from recent To face information in, filter out with the higher face information of verification sample matches degree, carry out identity examination ＆ verification and identity Mark, such as, the face information of certain employee to collecting is audited, and the face information that passes through of examination ＆ verification is marked it is upper its The identity such as name, department and job number, then, corresponding verification sample is updated further according to the mark, with after an action of the bowels The continuous verification sample that can be used after the renewal face of the employee be identified, etc..

In the research and practice process to prior art, it was found by the inventors of the present invention that due to existing scheme needs pair Face information carries out manual examination and verification and mark, therefore, it is necessary to expend more label time, causes treatment effeciency relatively low, moreover, Easily there is maloperation, influence recognition effect.

The content of the invention

The embodiment of the present invention provides a kind of face information mask method, device and storage medium；Mark can not only be improved Treatment effeciency, furthermore, it is possible to avoid maloperation, improve recognition effect.

The embodiment of the present invention provides a kind of face information mask method, including：

Obtain face information to be identified；

The face information to be identified is identified according to default verification sample set, is identified result, the knowledge Other result includes destination object and the face information to be identified and the matching degree of the destination object, the target pair As verifying sample with the face information matching degree highest to be identified for described verify in sample set；

If the face information to be identified and the matching degree of destination object meet preparatory condition, to the people to be identified Face information is labeled, face information after being marked；

The destination object in verification sample set is updated according to face information after the mark.

Accordingly, the embodiment of the present invention also provides a kind of face information annotation equipment, including：

Acquiring unit, for obtaining face information to be identified；

Recognition unit, for the face information to be identified to be identified according to default verification sample set, known Other result, the recognition result include destination object and the face information to be identified and the matching journey of the destination object Degree, the destination object verify sample for described verify in sample set with the face information matching degree highest to be identified；

Unit is marked, for when the matching degree of the face information to be identified and destination object meets preparatory condition, The face information to be identified is labeled, face information after being marked；

Updating block, for being carried out more to the destination object in verification sample set according to face information after the mark Newly.

In addition, the embodiment of the present invention also provides a kind of storage medium, the storage medium is stored with a plurality of instruction, the finger Order is loaded suitable for processor, and the step in the face information mask method described in 1 to 8 any one is required with perform claim.

Face information to be identified is identified according to default verification sample set for the embodiment of the present invention, is identified tying After fruit, (it can verify in sample set and be treated with this according to face information to be identified indicated in recognition result and destination object Identify face information matching degree highest verification sample) matching degree, automatic marking is carried out to face information to be identified, And the destination object in verification sample set is updated according to face information after mark, after subsequently can be with the renewal Destination object be foundation, face is identified；Because the program can be audited and marked automatically, and to verifying sample The verification sample of concentration is updated, for the existing scheme that manually can only be audited and marked, Ke Yi great It is big to improve the treatment effeciency of mark, moreover, it is also possible to avoid the maloperation caused by manual operation, improve verification sample Accuracy, and then be advantageous to improve the recognition effect of recognition of face.

Brief description of the drawings

Technical scheme in order to illustrate the embodiments of the present invention more clearly, make required in being described below to embodiment Accompanying drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the present invention, for For those skilled in the art, on the premise of not paying creative work, it can also be obtained according to these accompanying drawings other attached Figure.

Fig. 1 a are the schematic diagram of a scenario of face information mask method provided in an embodiment of the present invention；

Fig. 1 b are the flow charts of face information mask method provided in an embodiment of the present invention

Fig. 2 a are the schematic diagram of a scenario of face information labeling system provided in an embodiment of the present invention；

Fig. 2 b are the frame diagrams of face information mask method provided in an embodiment of the present invention；

Fig. 2 c are another flow charts of face information mask method provided in an embodiment of the present invention；

Fig. 3 a are the structural representations of face information annotation equipment provided in an embodiment of the present invention；

Fig. 3 b are another structural representations of face information annotation equipment provided in an embodiment of the present invention；

Fig. 4 is the structural representation of the network equipment provided in an embodiment of the present invention.

Embodiment

Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on Embodiment in the present invention, the every other implementation that those skilled in the art are obtained under the premise of creative work is not made Example, belongs to the scope of protection of the invention.

The embodiment of the present invention provides a kind of face information mask method, device and storage medium.Wherein, the face information mark Dispensing device can be specifically integrated in the network equipments such as server or server cluster.

For example, so that the face information annotation equipment specifically integrates in the server as an example, referring to Fig. 1 a, collecting device is being adopted After collecting image or video, corresponding face information to be identified can be extracted from the image or video collected, then, will The face information to be identified is sent to server, and the face information to be identified is entered according to default verification sample set by server Row identification, wherein, recognition result can include destination object and the face information to be identified and destination object (the i.e. school Test in sample set with face information matching degree highest to be identified verification sample) the information such as matching degree, hereafter, service Device can meet matching degree the face information to be identified of preparatory condition is labeled, and according to face information pair after the mark The destination object in verification sample set is updated, so as to subsequently can be using the destination object after the renewal as foundation, to people Face is identified, so as to realize that the verification sample in verification sample set can automatically update, to improve the purpose of recognition effect.

It is described in detail individually below.

Embodiment one,

The present embodiment will be described from the angle of face information annotation equipment, and the face information annotation equipment specifically can be with It is integrated in the network equipment, such as the equipment such as independent server or service cluster, wherein, the service cluster can include identification Equipment, cloud storage service device, annotation server and training server etc..

A kind of face information mask method, including：Face information to be identified is obtained, according to default verification sample set to this Face information to be identified is identified, and is identified result, and the recognition result includes destination object and the face letter to be identified Breath and the matching degree of the destination object, the destination object be in the verification sample set with the face information matching degree to be identified Highest verifies sample, if the face information to be identified and the matching degree of destination object meet preparatory condition, this is waited to know Other face information is labeled, face information after being marked, according to face information after the mark to being somebody's turn to do in verification sample set Destination object is updated.

As shown in Figure 1 b, the face information mask method can also include：

101st, face information to be identified is obtained.

For example, specifically image or video can be gathered by acquisition component, then carried from the image or video collected Take the face information to be identified；Or the image or video of collecting device transmission can also be received, then, from the figure collected The face information to be identified is extracted in picture or video；Or the face information to be identified of collecting device transmission can also be received, Wherein, the face information to be identified is extracted by the collecting device and obtained from the image or video collected.

Wherein, the acquisition component can be specifically camera device, such as photographing module or scan module part, and gather and set Standby can be specifically terminal, such as camera, video camera, camera, mobile phone, tablet personal computer, notebook computer, and/or individual Computer (PC, Personal Computer) etc..

It should be noted that the mode that face information to be identified is extracted from image or video can have it is a variety of, such as, can be with By human face detection tech, such as reference template method, face rule method, sample learning method or complexion model method come from image or Face information to be identified is extracted in video, or, can also by face tracking technology or face alignment technology come from image or Face information to be identified, etc. is extracted in video, will not be repeated here.

102nd, the face information to be identified is identified according to default verification sample set, is identified result.

Wherein, the recognition result can include destination object and the face information to be identified and of the destination object With information such as degree, the destination object is to verify sample with the face information matching degree highest to be identified in the verification sample set This.

Wherein, the default verification sample set can include multiple verification samples, specifically can be according to the need of practical application Ask and be configured；For example, so that the verification sample set is specially registry as an example, the registry can be used for storing each application system The face information of known registered user in system, such as, in current gate scene, user can be in advance in current gate system In registered to ask to be passed through based on face recognition technology, then registry is then for storing the people of these registered users Face information；For another example, in gate inhibition's scene, user can be registered in gate control system to ask to use recognition of face in advance Function control gate inhibition, then registry is then used for the face information for storing these registered users；Etc..The verification sample set is (such as Registry) it can both be positioned in the server of each the Internet, applications scene, it may be alternatively stored in other equipment such as cloud storage service In device, and it is supplied to the server of each the Internet, applications scene to be downloaded or call, etc..

It should be noted that initial verification sample can be configured in advance by user or attendant, such as, Ke Yi During user's registration, the image of user is gathered by acquisition component or collecting device, therefrom extracts the face characteristic information of user, And the identity of upper relative users is marked to the face characteristic information, then, save as verification sample corresponding to the identity This, etc.；Hereafter, in order to ensure the validity of the verification sample in the verification sample set, and the effect of raising recognition of face, In real time or regularly the verification sample in the verification sample set can be updated, specific update method can refer to step 104, It wouldn't repeat herein.

Wherein, can be had according to default verification sample set mode that the face information to be identified is identified it is a variety of, For example, specifically can be as follows：

(1) human face recognition model is obtained.

For example the human face recognition model can be specifically obtained from local (i.e. the face information annotation equipment), or, The human face recognition model can be obtained from other storage devices.

Wherein, the human face recognition model can be established in advance by attendant, or, can also be by the face information Annotation equipment is voluntarily established, i.e., before step " acquisition human face recognition model ", the face information mask method can be with Including：

Recognition of face initial model and face database are obtained, it is initial to the recognition of face according to the face database Model is trained, and obtains human face recognition model.

Such as specifically can be according to the face database, using higher-dimension local binary patterns (LBP, Local Binary Patterns) algorithm, principal component analysis (PCA, principal components analysis) algorithm, linear discriminant analysis (LDA, Linear Discriminate Analysis) joint bayesian algorithm, metric learning algorithm, transfer learning algorithm or Deep neural network scheduling algorithm is trained to the recognition of face initial model, obtains human face recognition model, etc., herein no longer Repeat.

Wherein, the recognition of face initial model can be advance by developer or attendant according to the demand of practical application Established, and the face database can include multiple face samples, the face sample can be from default face database In, or acquisition, etc. from internet.

(2) by the human face recognition model, by the face information to be identified and the verification sample in default verification sample set Matched respectively, obtain the face information to be identified and the matching degree of each verification sample；For example, specifically can be as follows：

By the human face recognition model, face characteristic extraction is carried out to the face information to be identified, obtains fisrt feature letter Breath；By the human face recognition model, face characteristic extraction is carried out to the verification sample in the verification sample set, obtains each verification Second feature information corresponding to sample；Fisrt feature information second feature information corresponding with each verification sample is carried out respectively Matching, obtain the face information to be identified and the matching degree of each verification sample.

Wherein, the matching degree can be embodied by modes such as similarity, scoring or match grades.For example, with similar Exemplified by degree, then now, fisrt feature information second feature information corresponding with each verification sample can be specifically calculated respectively Similarity, using the similarity being calculated as the face information to be identified and the matching degree of corresponding verification sample, such as, the The similarity of one characteristic information second feature information corresponding with verification sample X, you can as the face information to be identified and school Sample X matching degree is tested, by that analogy, etc..

Can be that a range of similarity sets score value corresponding to one accordingly in another example by taking scoring as an example, such as, If similarity is more than 90%, score value is 10 points, if between similarity is 80% to 90%, score value is 9 points, if similarity Between 70% to 80%, then score value is 8 points, if between similarity is 60% to 70%, score value is 7 points, by that analogy, etc. Deng.

Optionally, specific score value can also be not provided with, but is corresponding matching of a range of similarity division etc. Level, such as, if similarity is more than 90%, match grade is " height ", if between similarity is 60% to 90%, matching etc. Level for " in ", if similarity is less than 60%, match grade is " low ", etc..

Wherein, depending on the specific division of score value and match grade can be with the demand of practical application, will not be repeated here.

(3) matching degree highest verification sample is defined as the destination object of the face information to be identified, and by the mesh Object and the face information to be identified are marked with the matching degree of the destination object as recognition result.

If for example, the face information to be identified and verification sample X1 matching degree are 10 points, with verifying sample X2 Be 7 points with degree, be 3 points with test samples X3 matching degree, then at this point it is possible to will verification sample X1 to be defined as this to be identified The destination object of face information, and by the destination object and the face information to be identified and the matching degree of the destination object (i.e. 10 points) are used as recognition result.

It should be noted that if multiple matching degree highests verification samples be present, at this point it is possible to from the plurality of matching journey A destination object as the face information to be identified is randomly selected in degree highest verification sample, or, can also be according to Other Selection Strategies verify from the plurality of matching degree highest and conduct face to be identified letter are randomly selected in sample Destination object of breath, etc., it will not be repeated here.

If the 103rd, the face information to be identified and the matching degree of destination object meet preparatory condition, to the people to be identified Face information is labeled, face information after being marked.

If for example, the face information to be identified and the matching degree of destination object are higher than predetermined threshold value, the target is obtained The identity of object, for the identity of the upper destination object of face information mark to be identified, face is believed after being marked Breath.

Wherein, the identity can include name, account number, numbering, job number and/or the ID card No. of the destination object Etc. information, in addition, it can include other remark informations, such as department information, contact method and/or job information etc..

Optionally, it is labeled to the face information to be identified, can also should after being marked after face information Face information is concentrated added to labeled data after mark, and according to this, labeled data set pair human face recognition model has been trained, To be updated to the human face recognition model.

104th, the destination object in verification sample set is updated according to face information after the mark, subsequently may be used To carry out recognition of face using the verification sample set after the renewal.

Wherein, the mode of renewal can have a variety of, for example, can specifically use any one following mode：

(1) first way；

The destination object verified in sample set is replaced with into face information after the mark.

For example if registered user A destination objects in sample set is verified are " verification sample 1 ", at this point it is possible to verify In sample set, verification sample 1 is replaced with to face information, etc. after marking.

(2) second way；

Calculate face information and the average of the destination object after the mark according to preset algorithm, will verify in sample set should Destination object is replaced with the average.

Wherein, the preset algorithm can be configured according to the demand of practical application, such as, it can directly calculate the mark Face information and the average of the destination object afterwards；Or face information and the destination object multiply respectively after can also this be marked With corresponding weight, after the mark after being weighted face information and weighting after destination object, then, then calculate weighting after The average of destination object, etc. after mark after face information and weighting, will not be repeated here.

From the foregoing, it will be observed that face information to be identified is identified according to default verification sample set for the present embodiment, obtain After recognition result, (it can be verified according to face information to be identified indicated in recognition result and destination object in sample set With face information matching degree highest to be identified verification sample) matching degree, face information to be identified is carried out from Dynamic mark, and being updated according to face information after mark to the destination object in verification sample set, so as to subsequently can be with Destination object after the renewal is foundation, and face is identified；Because the program can be audited and marked automatically, and it is right Verification sample in verification sample set is updated, accordingly, with respect to the existing scheme that manually can only be audited and marked Speech, can greatly improve the treatment effeciency of mark, moreover, it is also possible to avoid the maloperation caused by manual operation, improve The accuracy of sample is verified, and then is advantageous to improve the recognition effect of recognition of face.

Embodiment two,

According to the method described by upper one embodiment, citing is described in further detail below.

In the present embodiment, will be illustrated so that the face information annotation equipment is specifically integrated in server cluster as an example. Wherein, the server cluster can include multiple different equipment, and the plurality of equipment can be performed in unison with face information mark Step in method.It should be noted that the function distribution in the server cluster between each equipment can be according to practical application Demand be configured, such as, face information to be identified can be identified by identification equipment, by cloud storage service device Lai Data are preserved, face information to be identified is labeled by annotation server, entered by training server come human face recognition model Row training, etc., will be described in more detail below.

As shown in Figure 2 a, the figure is a schematic diagram of a scenario of face identification system.The face identification system is except can be with Outside server cluster, other equipment, such as collecting device and application apparatus etc. can also be included, wherein, Ge Geshe Standby function specifically can be as follows：

(1) collecting device；

Collecting device, for gathering image or video, and face to be identified is extracted from the image or video collected Information, face information to be identified is supplied to identification equipment.

Wherein, it is terminal that collecting device, which is specifically as follows, for example, camera, video camera, camera, mobile phone, tablet personal computer, Notebook computer, and/or PC etc..

(2) server cluster；

The server cluster can include multiple equipment, for convenience, in the present embodiment, will be with the server set Group includes illustrating exemplified by the equipment such as identification equipment, cloud storage service device, annotation server and training server, wherein, respectively The function of individual equipment specifically can be as follows：

A, identification equipment；

Identification equipment, for receiving the face information to be identified of collecting device transmission, obtain face from training server and know Other model, and verification sample set is obtained from cloud storage service device, then, using the human face recognition model, according to the verification sample This set pair face information to be identified is identified, and is identified result, wherein, the recognition result include destination object and The face information to be identified and the matching degree of the destination object, then, on the one hand, recognition result is sent to application apparatus, So that application apparatus is handled according to the recognition result, on the other hand, by the face information to be identified and corresponding identification As a result cloud storage service device is sent to, to be preserved.

Wherein, so-called destination object, refer in the verification sample set, with the face information matching degree to be identified most High verification sample.It should be noted that if multiple matching degree highests verification samples be present, at this point it is possible to from the plurality of A destination object as the face information to be identified is randomly selected in matching degree highest verification sample, or, also may be used It is to be identified as this to randomly select one from the plurality of matching degree highest verification sample according to other Selection Strategies Destination object of face information, etc., it will not be repeated here.

It should be noted that the identification equipment in addition to it can be laid out in server side, can also be installed in the terminal, or Person, it can also be integrated in same entity, will not be repeated here with collecting device.

B, cloud storage service device；

Cloud storage service device, sample set is verified for preserving, verification sample set is supplied to identification equipment, and receive identification The face information to be identified and the corresponding recognition result of face information to be identified that equipment returns, are preserved and this is to be identified Face information and the corresponding recognition result of face information to be identified are sent to annotation server；Hereafter, mark can also be received Face information after the mark that note server returns, the destination object in verification sample set is entered according to face information after the mark Row renewal.

Optionally, cloud storage service device, can be also used for by this mark after face information added to labeled data concentrate, And labeled data collection is supplied to training server, to carry out the training of model.

Optionally, in order to save Internet resources, labeled data collection is being supplied to training server by cloud storage service device When, except that only labeled data can also be concentrated in addition to entirely labeled data collection is supplied to training server new The face information of mark increased is supplied to training server, will not be repeated here.

C, annotation server；

Annotation server, for receiving the face information to be identified and the face to be identified of the transmission of cloud storage service device The corresponding recognition result of information, wherein, the recognition result includes destination object and the face information to be identified and destination object Matching degree, if the face information to be identified and the matching degree of destination object meet preparatory condition, to the people to be identified Face information is labeled, face information after being marked, and face information after mark is sent into cloud storage service device, to be protected Deposit.

D, training server；

Training server, for obtaining recognition of face initial model and face database, according to the face database pair The recognition of face initial model is trained, and obtains human face recognition model, and the human face recognition model is sent into identification equipment.

Optionally, the training server, can be also used for obtaining labeled data collection from cloud storage server, according to this Labeled data set pair human face recognition model is trained, to be updated to the human face recognition model.

(3) application apparatus；

Application apparatus, for receiving the recognition result of identification equipment transmission, predetermined registration operation is carried out according to the recognition result, than Such as, control or punched-card machine operation of access control, etc. are carried out.

Based on the framework of above-mentioned face identification system, flow will be performed to it below and be described in detail.

As shown in Figure 2 b, a kind of face information mask method, idiographic flow can be as follows：

201st, collecting device collection image or video, and face letter to be identified is extracted from the image or video collected Breath, identification equipment is supplied to by face information to be identified.

For example, can specifically monitor the user to be identified for entering to image acquisition region by collecting device, and this is waited to know Other user is taken pictures or recorded a video, and obtains the image or video on the user to be identified, then, from the image collected or is regarded Face information to be identified is extracted in frequency, face information to be identified is supplied to identification equipment, etc..

Wherein, image acquisition region can include the request region of current gate, the request region of gate inhibition, self-help certificate handling Operating area of operating area, and/or identification authentication etc., and it is terminal that collecting device, which is then specifically as follows, for example camera, take the photograph Camera, camera, mobile phone, tablet personal computer, notebook computer, and/or PC etc..

It should be noted that the image collected or video can also be sent to identification equipment by collecting device, set by identification It is standby to extract face information to be identified from the image or video received, for convenience, the present embodiment will with by Collecting device illustrates exemplified by extracting face information to be identified, will not be repeated here.

Wherein, extracted from image or video the mode of face information to be identified can have it is a variety of, such as, people can be passed through Face detection tech, such as reference template method, face rule method, sample learning method or complexion model method are come from image or video Face information to be identified is extracted, or, can also be by face tracking technology or face alignment technology come from image or video Face information to be identified, etc. is extracted, is not limited thereto.

202nd, identification equipment obtains after the face information to be identified of collecting device transmission is received from cloud storage service device Verify sample set.

Wherein, the default verification sample set can include multiple verification samples, specifically can be according to the need of practical application Ask and be configured.

It should be noted that initial verification sample can be configured in advance by user or attendant, such as, Ke Yi During user's registration, the image of user is gathered by acquisition component or collecting device, therefrom extracts the face characteristic information of user, And the identity of upper relative users is marked to the face characteristic information, then, save as verification sample corresponding to the identity This, etc..

For example, the structure of the verification sample set can be as described in table one.

Table one：

It should be noted that above-mentioned table one is only for example, the verification sample set can also use the storage of other non-forms Mode, such as array etc.；In addition, each item data in table one is only for example；If it is understood that only wrapped in user images Include piece image, then corresponding face characteristic information can be directly calculated by the image；If user images include one Width images above, then what the characteristic value that corresponding face characteristic information can be then calculated by each image obtained after being averaging Value, will not be repeated here.

Hereafter, in order to ensure the validity of the verification sample in the verification sample set, and the effect of raising recognition of face, In real time or regularly the verification sample in the verification sample set can be updated, such as, reference can be made to step 211, herein wouldn't Repeat.

203rd, identification equipment obtains human face recognition model from training server.

Wherein, the human face recognition model can be established in advance by attendant, or, can also be by the face information Annotation equipment is voluntarily established, i.e., before step " identification equipment obtains human face recognition model from training server ", the people Face information labeling method can also include：

Training server obtains recognition of face initial model and face database, according to the face database to the people Face identification initial model is trained, and obtains human face recognition model.

For example training server specifically can combine shellfish according to the face database using LBP algorithms, PCA algorithms, LDA Leaf this algorithm, metric learning algorithm, transfer learning algorithm or deep neural network scheduling algorithm enter to the recognition of face initial model Row training, obtains human face recognition model, etc., will not be repeated here.

Wherein, the recognition of face initial model can be advance by developer or attendant according to the demand of practical application Established, and the face database can include multiple face samples, the face sample can be from default face database In, or obtained from internet, for example can be obtained from internet in many disclosed face databases, or it is real from history Trample and gather and obtained in the history face database that accumulates the human face data of user and formed, or in advance will executor Collecting device is set up at the application scenarios of face identification, the personnel that arrange work gather the human face datas of some known users to build in advance The face database is found, then obtained from the face database of the foundation, etc..

It should be noted that step 202 and 203 execution can be in no particular order.

204th, identification equipment is verified the face information to be identified in sample set with default by the human face recognition model Verification sample is matched respectively, obtains the face information to be identified and the matching degree of each verification sample；For example, with reference to figure 2c, specifically can be as follows：

(1) identification equipment is carried out face characteristic extraction to the face information to be identified, obtained by the human face recognition model Fisrt feature information.

Wherein, face characteristic refers to can be used for each critical zone locations of positioning face, for example, eyebrow, eyes, nose, The information of the position at the position such as face and face mask, such as, specifically can include 2 eyeball central points, 4 canthus points, The midpoint in two nostrils and 2 corners of the mouth points, etc., will not be repeated here.

(2) identification equipment carries out face characteristic by the human face recognition model to the verification sample in the verification sample set Extraction, obtain second feature information corresponding to each verification sample.

For example, include verifying sample 1, verification sample 2, verification sample 3 and verification sample with the verification sample Exemplified by 4 ... ..., then identification equipment specifically can be by the human face recognition model, respectively to verification sample 1, verification sample 2, school Test sample 3 and verification sample 4 ... etc. and carry out face characteristic extraction, respectively obtain second corresponding to these verification samples Characteristic information, it for details, reference can be made to table two.

Table two：

Verify sample	Second feature information	……
			Verify sample 1	Second feature information 1	……
Verify sample 2	Second feature information 2	……
			Verify sample 3	Second feature information 3
Verify sample 4	Second feature information 4
			……	……	……

As shown in Table 2, the second feature information of sample 1 is verified as " second feature information 1 ", the second of verification sample 2 are special Reference breath is " second feature information 2 ", the second feature information of verification sample 3 are " second feature information 3 ", to verify sample 4 Second feature information is " second feature information 4 ", etc..

It should be noted that if in sample set is verified, if the face characteristic information for having saved each verification sample, Then at this time it is also possible to face characteristic information corresponding to sample be verified needed for directly being obtained from verification sample set, without leading to again The human face recognition model is crossed, face characteristic extraction is carried out to the verification sample in the verification sample set.

(3) identification equipment carries out fisrt feature information and the corresponding second feature information of each verification sample respectively Match somebody with somebody, obtain the face information to be identified and the matching degree of each verification sample, such as, reference can be made to Fig. 2 c.

If for example, in step (2), the second feature information for obtaining verifying sample 1 is " second feature information 1 ", verification The second feature information of sample 2 is " second feature information 2 ", the second feature information of verification sample 3 is " second feature information 3 ", the second feature information of verification sample 4 is " second feature information 4 ", then at this point it is possible to perform following operation：

By fisrt feature information with " second feature information 1 " is compared, obtain the face information to be identified with verification sample This 1 matching degree；

By fisrt feature information with " second feature information 2 " is compared, obtain the face information to be identified with verification sample This 2 matching degree；

By fisrt feature information with " compared with second feature information 3 ", obtaining the face information to be identified and verification sample This 3 matching degree；

By fisrt feature information with " second feature information 4 " is compared, obtain the face information to be identified with verification sample This 4 matching degree；

By that analogy, etc..

Wherein, the matching degree can be embodied by modes such as similarity, scoring or match grades.For example, with similar Exemplified by degree, then now, fisrt feature information second feature information corresponding with each verification sample can be specifically calculated respectively Similarity, using the similarity being calculated as the face information to be identified and the matching degree of corresponding verification sample, such as, the One characteristic information and the similarity of second feature information 1, you can as the face information to be identified and the matching journey of verification sample 1 Degree, the similarity of fisrt feature information and second feature information 2, you can as the face information to be identified and verification sample 2 Matching degree, by that analogy, etc..

Can be that a range of similarity sets score value corresponding to one accordingly in another example by taking scoring as an example, such as, Using similarity as more than 90%, corresponding score value is that full marks 5 divide, and between similarity is 70% to 90%, corresponding score value is 4 points, phase Between being 50% to 70% like degree, corresponding score value is 3 points, and between similarity is 30% to 50%, corresponding score value is 2 points, similar Spend between 10% to 30%, corresponding score value is 1 point, between similarity is 0% to 10%, it is that example is carried out that corresponding score value, which is 0 point, Illustrate, if then the similarity of fisrt feature information and second feature information 1 is 85%, then, can be to determine the people to be identified Face information and the matching degree of verification sample 1 are 5 points, similarly, if the similarity of fisrt feature information and second feature information 2 is 53%, the similarity of fisrt feature information and two characteristic informations 3 for 18% so, can with determine the face information to be identified with The matching degree for verifying sample 1 is 3 points, and determines the face information to be identified with verifying the matching degree of sample 3 as 1 point, By that analogy, etc..

205th, matching degree highest verification sample is defined as the destination object of the face information to be identified by identification equipment, And using the destination object and the face information to be identified and the matching degree of the destination object as recognition result.

If for example, the face information to be identified with verification sample 1 matching degree be 5 points, with verify sample 2 matching journey Spend for 3 points, the matching degree with test samples 3 is 1 point, then believes at this point it is possible to which verification sample 1 is defined as into the face to be identified The destination object of breath, and by the destination object and the face information to be identified and the matching degree (i.e. 5 points) of the destination object As recognition result.

For example if the identity of verification sample 1 is that " Finance Department, Zhang San, job number 00001 " then now, can be with determination The identity of the face information owning user to be identified for " Finance Department, Zhang San, job number 00001 ", by that analogy, etc..

206th, the face information to be identified and corresponding recognition result are sent to cloud storage service device by identification equipment.

If for example, in step 205, sample 1 will be verified it is defined as the destination object of the face information to be identified, and should Destination object and the face information to be identified and the matching degree of the destination object, i.e., 5 are allocated as recognition result, then now, The face information to be identified, verification sample 1 and " 5 points " can be sent to cloud storage service device.

207th, cloud storage service device is treated after the face information to be identified and corresponding recognition result is received to this Identification face information and corresponding recognition result are preserved, and the face information to be identified and corresponding identification are tied Fruit is sent to annotation server.

208th, server is annotated after the face information to be identified and corresponding recognition result is received, it is determined that this is treated Whether identification face information and the matching degree of destination object meet preparatory condition, if so, step 209 is then performed, if it is not, then neglecting The slightly face information to be identified and corresponding recognition result.

For example, annotation service implement body can determine whether the matching degree of the face information to be identified and destination object is high In predetermined threshold value, if so, step 209 is then performed, if it is not, then ignoring the face information to be identified and corresponding recognition result.

Optionally, when the face information to be identified and the matching degree of destination object are less than or equal to predetermined threshold value, except It can directly ignore outside the face information to be identified, other processing modes can also be used, such as, can be to be identified by this Face information is added in verification sample set, and manually marks upper corresponding identity, etc..

Wherein, the predetermined threshold value can be configured according to the demand of practical application, such as, or it is specific with matching degree It is presented as " scoring ", and it is example that the predetermined threshold value, which is 4 points, because the face information to be identified and destination object (verify sample 1) matching degree is " 5 points ", so, it can now determine that the matching degree meets preparatory condition, and then step can be performed 209。

Optionally, when it is determined that the face information to be identified and the matching degree of destination object are unsatisfactory for preparatory condition, also Corresponding prompt message can be generated, it is pre- to remind the matching degree of user's face information to be identified and destination object to be unsatisfactory for If condition, specific suggestion content can be configured according to the demand of practical application, will not be repeated here.

209th, server is annotated it is determined that the face information to be identified and the matching degree of destination object meet preparatory condition When, the face information to be identified is labeled, face information after being marked.

For example, with reference to Fig. 2 c, if the face information to be identified and the matching degree of destination object are higher than predetermined threshold value, obtain The identity of the destination object is taken, for the identity of the upper destination object of face information mark to be identified, is marked Face information afterwards.

Such as by destination object for " verification sample 1 " exemplified by, if verification sample 1 identity for " Finance Department, Zhang San, Job number 00001 ", then now, can think the face information to be identified mark " Finance Department, Zhang San, job number 00001 ", with This analogizes, etc..

210th, annotate server by this mark after face information be sent to cloud storage service device.

211st, cloud storage service device, can be according to face information pair after the mark after face information after receiving the mark The destination object in verification sample set is updated.

For example, specifically directly the destination object in verification sample set can be carried out more according to face information after mark Newly, or, as shown in Figure 2 c, can also first by this mark after face information added to labeled data concentrate, then, deposited by cloud The destination object of storage server in the sample set of labeled data set pair verification is updated, etc..

(1) first way；

The destination object verified in sample set is replaced with face information after the mark by cloud storage service device.

Such as if registered user " Zhang San " in sample set is verified it is original verification sample for " verification sample 1 ", this When, verification sample 1 can be replaced with into face information, etc. after the mark in sample set is verified.

(2) second way；

Cloud storage service device calculates face information and the average of the destination object after the mark according to preset algorithm, will verify The destination object in sample set is replaced with the average.

Such as or by taking registered user " Zhang San " as an example, if registered user " Zhang San " original school in sample set is verified Sample is tested as " verification sample 1 ", then at this point it is possible to calculate " verification sample 1 " and " face information after mark " according to preset algorithm Average, then, in the verification sample set, will verification sample 1 replace with the average, etc..

Optionally, as shown in Figure 2 c, face information is added to and has marked number after cloud storage service device can also mark this According to concentration, then, by cloud storage service device, by this, labeled data collection has been supplied to training server, by training server according to this Labeled data set pair human face recognition model is trained, to be updated to the human face recognition model, specific update method Reference can be made to upper one embodiment, will not be repeated here.

From the foregoing, it will be observed that face information to be identified is identified according to default verification sample set for the present embodiment, obtain After recognition result, (it can be verified according to face information to be identified indicated in recognition result and destination object in sample set With face information matching degree highest to be identified verification sample) matching degree, face information to be identified is carried out from Dynamic mark, and being updated according to face information after mark to the destination object in verification sample set, so as to subsequently can be with Destination object after the renewal is foundation, and face is identified；Because the program can be audited and marked automatically, and it is right Verification sample in verification sample set is updated, accordingly, with respect to the existing scheme that manually can only be audited and marked Speech, can be greatly lowered mark cost, greatly improve the treatment effeciency of mark, and improve the feasibility and easily of labeling operation Operability, further, it is also possible to avoid the maloperation caused by manual operation, improve the accuracy of verification sample, Jin Eryou Beneficial to the recognition effect for improving recognition of face.

Further, since the face information of the user of actual acquisition can be from each side such as scene, angle, light, definition Face and usage scenario reach maximum fitting, therefore, the data and reality in verification sample set after being updated accordingly Data can tend to " homologous ", so, even if user profile gradually changes, the verification sample verified in sample set can also be protected It is similar to active user to hold its corresponding face characteristic information, the face identification system can be avoided to produce obvious performance and moved back Change, improve recognition performance；The time being additionally, since spent by renewal verification sample set is shorter, and real-time is more excellent, therefore, it is possible to Situation is actually used according to user to be adjusted rapidly, is advantageous to be lifted the usage experience of user

Embodiment three,

In order to preferably implement above method, the embodiment of the present invention also provides a kind of face information annotation equipment, such as Fig. 3 a Shown, it is single that the face information annotation equipment can include acquiring unit 301, recognition unit 302, mark unit 303 and renewal Member 304, specifically can be as follows：

(1) acquiring unit 301；

Acquiring unit 301, for obtaining face information to be identified.

For example, acquiring unit 301, specifically can gather image or video, then from the figure collected by acquisition component The face information to be identified is extracted in picture or video；Or the image or video of collecting device transmission can also be received, then, The face information to be identified is extracted from the image or video collected；Or treating for collecting device transmission can also be received Face information is identified, wherein, the face information to be identified is extracted by the collecting device and obtained from the image or video collected Arrive.

Wherein, the acquisition component can be specifically camera device, such as photographing module or scan module part, and gather and set Standby can be specifically terminal, such as camera, video camera, camera, mobile phone, tablet personal computer, notebook computer, and/or PC etc..

(2) recognition unit 302；

Recognition unit 302, for the face information to be identified to be identified according to default verification sample set, known Other result.

Wherein, the default verification sample set can include multiple verification samples, specifically can be according to the need of practical application Ask and be configured；It should be noted that initial verification sample can be configured in advance by user or attendant, such as, can So that in user's registration, the image of user is gathered by acquisition component or collecting device, the face characteristic of user is therefrom extracted Information, and the identity of upper relative users is marked to the face characteristic information, then, save as school corresponding to the identity Test sample, etc..

Wherein, can be had according to default verification sample set mode that the face information to be identified is identified it is a variety of, For example, the recognition unit 302 can include obtaining subelement, coupling subelement and determination subelement, it is as follows：

Subelement is obtained, can be used for obtaining human face recognition model.

For example acquisition subelement can specifically obtain the recognition of face mould from local (i.e. the face information annotation equipment) Type, or, the human face recognition model can also be obtained from other storage devices.

Coupling subelement, it can be used for by the human face recognition model, by the face information to be identified and default verification sample The verification sample of this concentration is matched respectively, obtains the face information to be identified and the matching degree of each verification sample.

For example, the coupling subelement, specifically can be used for：By the human face recognition model, to the face information to be identified Face characteristic extraction is carried out, obtains fisrt feature information；By the human face recognition model, to the verification sample in the verification sample set This progress face characteristic extraction, obtain second feature information corresponding to each verification sample；By fisrt feature information and each school Test second feature information corresponding to sample to be matched respectively, obtain the face information to be identified and the matching of each verification sample Degree.

Wherein, the matching degree can be embodied by modes such as similarity, scoring or match grades, before for details, reference can be made to The embodiment of the method in face, will not be repeated here.

Determination subelement, it can be used for the mesh that matching degree highest verification sample is defined as to the face information to be identified Object is marked, and using the destination object and the face information to be identified and the matching degree of the destination object as recognition result.

It should be noted that if multiple matching degree highest verification samples be present, now, determination subelement can be from this A destination object as the face information to be identified is randomly selected in multiple matching degree highest verification samples, or, It can also be verified according to other Selection Strategies from the plurality of matching degree highest in sample and randomly select a conduct this is treated Destination object of face information, etc. is identified, will not be repeated here.

Wherein, human face recognition model can be established in advance by attendant, or, can also be by the face information mark Dispensing device is voluntarily established, such as, as shown in Figure 3 b, the face information annotation equipment can be as follows with training unit 306：

Training unit 306, it can be used for obtaining recognition of face initial model and face database, according to the face number The recognition of face initial model is trained according to storehouse, obtains human face recognition model.

(3) unit 303 is marked；

Unit 303 is marked, for when the matching degree of the face information to be identified and destination object meets preparatory condition, The face information to be identified is labeled, face information after being marked.

For example, the mark unit, specifically can be used for high in the face information to be identified and the matching degree of destination object When predetermined threshold value, the identity of the destination object is obtained, for the body of the upper destination object of face information mark to be identified Part mark, face information after being marked.

(4) updating block 304；

Updating block 304, for being carried out more to the destination object in verification sample set according to face information after the mark Newly.

So, follow-up recognition unit 302 can be so as to using the verification sample set progress recognition of face after the renewal.

The updating block 304, it specifically can be used for the destination object verified in sample set replacing with mark descendant Face information.

Or the updating block 304, it specifically can be used for calculating face information and the mesh after the mark according to preset algorithm The average of object is marked, the destination object verified in sample set is replaced with the average.

Optionally, it is labeled to the face information to be identified, can also should after being marked after face information Face information is concentrated added to labeled data after mark, and according to this, labeled data set pair human face recognition model has been trained, To be updated to the human face recognition model；I.e. as shown in Figure 3 b, the face information annotation equipment can also include adding device 305, it is as follows：

The adding device 305, concentrated for face information after this is marked added to labeled data；

Then now, training unit 306, can be also used for that labeled data set pair human face recognition model has been instructed according to this Practice, to be updated to the human face recognition model.

For example, the training unit 306, face information is carried out in advance after specifically can be used for the mark to the concentration of labeled data Processing, increment face information is obtained, obtain history training face information, history training face information is to have marked and had been used for The face information of human face recognition model training, training sample is determined according to the increment face information and history training face information, Human face recognition model is trained according to the training sample, to be updated to the human face recognition model.

Such as the training unit 306, it specifically can be used for obtaining history training face information list, the history training of human Face information list preserves history training face information, increment face information list is generated according to the increment face information, by this After the list of increment face information merges with history training face information list, by the suitable of the information in list after merging Sequence is ranked up according to preset rules, obtains training sample list, and human face recognition model is carried out according to the training sample list Training, to be updated to the human face recognition model.

It when it is implemented, above unit can be realized as independent entity, can also be combined, be made Realized for same or several entities, the specific implementation of above unit can be found in embodiment of the method above, herein not Repeat again.

From the foregoing, it will be observed that the face information annotation equipment of the present embodiment by recognition unit 302 according to default verification sample Set pair face information to be identified is identified, can be by mark unit 303 according to signified in recognition result after being identified result The face information to be identified shown and destination object (verify in sample set with the face information matching degree highest school to be identified Test sample) matching degree, to face information to be identified carry out automatic marking, and by updating block 304 according to mark descendant Face information to verification sample set in the destination object be updated, so as to subsequently can using the destination object after the renewal as according to According to face is identified；Because the program can be audited and marked automatically, and to the verification sample in verification sample set It is updated, for the existing scheme that manually can only be audited and marked, the place of mark can be greatly improved Efficiency is managed, moreover, it is also possible to avoid the maloperation caused by manual operation, improves the accuracy of verification sample, Jin Eryou Beneficial to the recognition effect for improving recognition of face.

Example IV,

The embodiment of the present invention also provides a kind of network equipment, as shown in figure 4, it illustrates involved by the embodiment of the present invention The structural representation of the network equipment, specifically：

The network equipment can include one or more than one processing core processor 401, one or more The parts such as memory 402, power supply 403 and the input block 404 of computer-readable recording medium.Those skilled in the art can manage Solve, the network equipment infrastructure shown in Fig. 4 does not form the restriction to the network equipment, can include more more or less than illustrating Part, either combine some parts or different parts arrangement.Wherein：

Processor 401 is the control centre of the network equipment, utilizes various interfaces and connection whole network equipment Various pieces, by running or performing the software program and/or module that are stored in memory 402, and call and be stored in Data in reservoir 402, the various functions and processing data of the network equipment are performed, so as to carry out integral monitoring to the network equipment. Optionally, processor 401 may include one or more processing cores；Preferably, processor 401 can integrate application processor and tune Demodulation processor processed, wherein, application processor mainly handles operating system, user interface and application program etc., and modulatedemodulate is mediated Reason device mainly handles radio communication.It is understood that above-mentioned modem processor can not also be integrated into processor 401 In.

Memory 402 can be used for storage software program and module, and processor 401 is stored in memory 402 by operation Software program and module, so as to perform various function application and data processing.Memory 402 can mainly include storage journey Sequence area and storage data field, wherein, storing program area can storage program area, the application program (ratio needed at least one function Such as sound-playing function, image player function) etc.；Storage data field can store uses created number according to the network equipment According to etc..In addition, memory 402 can include high-speed random access memory, nonvolatile memory can also be included, such as extremely Few a disk memory, flush memory device or other volatile solid-state parts.Correspondingly, memory 402 can also wrap Memory Controller is included, to provide access of the processor 401 to memory 402.

The network equipment also includes the power supply 403 to all parts power supply, it is preferred that power supply 403 can pass through power management System and processor 401 are logically contiguous, so as to realize management charging, electric discharge and power managed etc. by power-supply management system Function.Power supply 403 can also include one or more direct current or AC power, recharging system, power failure monitor The random component such as circuit, power supply changeover device or inverter, power supply status indicator.

The network equipment may also include input block 404, and the input block 404 can be used for the numeral or character for receiving input Information, and produce keyboard, mouse, action bars, optics or the trace ball signal relevant with user's setting and function control Input.

Although being not shown, the network equipment can also will not be repeated here including display unit etc..Specifically in the present embodiment In, the processor 401 in the network equipment can be corresponding by the process of one or more application program according to following instruction Executable file be loaded into memory 402, and the application program being stored in memory 402 is run by processor 401, It is as follows so as to realize various functions：

Face information to be identified is obtained, the face information to be identified is identified according to default verification sample set, obtained To recognition result, the recognition result includes destination object and the face information to be identified and the matching degree of the destination object, The destination object is to verify sample with the face information matching degree highest to be identified in the verification sample set, if this is to be identified Face information and the matching degree of destination object meet preparatory condition, then the face information to be identified are labeled, marked Face information after note, the destination object in verification sample set is updated according to face information after the mark.

For example, can specifically obtain human face recognition model, by the human face recognition model, the face information to be identified is entered Pedestrian's face feature extraction, obtains fisrt feature information；By the human face recognition model, to the verification sample in the verification sample set Face characteristic extraction is carried out, obtains second feature information corresponding to each verification sample；By fisrt feature information and each verification Second feature information is matched respectively corresponding to sample, obtains the face information to be identified and the matching journey of each verification sample Degree, then, matching degree highest verification sample is defined as the destination object of the face information to be identified, and by the target pair As and the face information to be identified and the destination object matching degree as recognition result.

The specific implementation of each operation can be found in embodiment above above, will not be repeated here.

From the foregoing, it will be observed that the network equipment of the present embodiment is carried out according to default verification sample set to face information to be identified Identification, after being identified result, (can it be verified with destination object according to face information to be identified indicated in recognition result In sample set with face information matching degree highest to be identified verification sample) matching degree, face to be identified is believed Breath carries out automatic marking, and the destination object in verification sample set is updated according to face information after mark, with after an action of the bowels It is continuous face to be identified using the destination object after the renewal as foundation；Due to the program can carry out automatically examination ＆ verification and Mark, and the verification sample in verification sample set is updated, manually it can only be audited and marked accordingly, with respect to existing Scheme for, the treatment effeciency of mark can be greatly improved, moreover, it is also possible to avoid the mistake caused by manual operation from grasping Make, improve the accuracy of verification sample, and then be advantageous to improve the recognition effect of recognition of face.

Embodiment five,

It will appreciated by the skilled person that all or part of step in the various methods of above-described embodiment can be with Completed by instructing, or control related hardware to complete by instructing, the instruction can be stored in one and computer-readable deposit In storage media, and loaded and performed by processor.

Therefore, the embodiment of the present invention provides a kind of storage medium, wherein being stored with a plurality of instruction, the instruction can be processed Device is loaded, to perform the step in any face information mask method that the embodiment of the present invention provided.For example, this refers to Order can perform following steps：

Optionally, the instruction can also carry out following steps：

It is labeled to the face information to be identified, after being marked after face information, after can also this be marked Face information is concentrated added to labeled data, and according to this, labeled data set pair human face recognition model has been trained, with to this Human face recognition model is updated.

Wherein, the storage medium can include：Read-only storage (ROM, Read Only Memory), random access memory Body (RAM, Random Access Memory), disk or CD etc..

By the instruction stored in the storage medium, any face letter that the embodiment of the present invention is provided can be performed The step in mask method is ceased, it is thereby achieved that any face information mask method institute that the embodiment of the present invention is provided The beneficial effect that can be realized, refers to embodiment above, will not be repeated here.

A kind of face information mask method, device and the storage medium provided above the embodiment of the present invention has been carried out in detail Thin to introduce, specific case used herein is set forth to the principle and embodiment of the present invention, and above example is said It is bright to be only intended to help the method and its core concept for understanding the present invention；Meanwhile for those skilled in the art, according to this hair Bright thought, there will be changes in specific embodiments and applications, in summary, this specification content should not manage Solve as limitation of the present invention.

Claims

A kind of 1. face information mask method, it is characterised in that including：

Obtain face information to be identified；

The face information to be identified is identified according to default verification sample set, is identified result, the identification knot Fruit includes destination object and the face information to be identified and the matching degree of the destination object, and the destination object is In the verification sample set sample is verified with the face information matching degree highest to be identified；

If the face information to be identified and the matching degree of destination object meet preparatory condition, the face to be identified is believed Breath is labeled, face information after being marked；

The destination object in verification sample set is updated according to face information after the mark.
2. according to the method for claim 1, it is characterised in that it is described according to it is default verification sample set to described to be identified Face information is identified, and is identified result, including：

Obtain human face recognition model；

By the human face recognition model, the face information to be identified and the verification sample in default verification sample set are distinguished Matched, obtain the face information to be identified and the matching degree of each verification sample；

Matching degree highest verification sample is defined as the destination object of the face information to be identified, and by the target pair As and the face information to be identified and the destination object matching degree as recognition result.
3. according to the method for claim 2, it is characterised in that it is described by the human face recognition model, wait to know by described Other face information is matched respectively with the verification sample in default verification sample set, obtain the face information to be identified with it is each The matching degree of individual verification sample, including：

By the human face recognition model, face characteristic extraction is carried out to the face information to be identified, obtains fisrt feature letter Breath；

By the human face recognition model, face characteristic extraction is carried out to the verification sample in the verification sample set, obtained each Second feature information corresponding to individual verification sample；

Fisrt feature information second feature information corresponding with each verification sample is matched respectively, obtained described to be identified Face information and the matching degree of each verification sample.
4. according to the method for claim 1, it is characterised in that if the face information to be identified and destination object Matching degree meets preparatory condition, then the face information to be identified is labeled, face information after being marked, including：

If the face information to be identified and the matching degree of destination object are higher than predetermined threshold value, the destination object is obtained Identity；

For the identity of the upper destination object of face information mark to be identified, face information after being marked.
5. according to the method described in any one of Claims 1-4, it is characterised in that described according to face information after the mark The destination object in verification sample set is updated, including：

The destination object verified in sample set is replaced with into face information after the mark.
6. according to the method described in any one of Claims 1-4, it is characterised in that described according to face information after the mark The destination object in verification sample set is updated, including：

According to face information and the average of the destination object after the preset algorithm calculating mark；

The destination object verified in sample set is replaced with the average.
7. according to the method for claim 2, it is characterised in that it is described that the face information to be identified is labeled, obtain After to mark after face information, in addition to：

Face information after the mark is concentrated added to labeled data；

It is trained according to the set pair human face recognition model of labeled data, to be updated to the human face recognition model.
8. according to the method for claim 2, it is characterised in that before the acquisition human face recognition model, in addition to：

Obtaining recognition of face initial model and face database, the face database includes multiple face samples；

The recognition of face initial model is trained according to the face database, obtains human face recognition model.
A kind of 9. face information annotation equipment, it is characterised in that including：

Acquiring unit, for obtaining face information to be identified；

Recognition unit, for the face information to be identified to be identified according to default verification sample set, it is identified tying Fruit, the recognition result include destination object and the face information to be identified and the matching degree of the destination object, institute Destination object is stated to verify sample with the face information matching degree highest to be identified in the verification sample set；

Unit is marked, for when the matching degree of the face information to be identified and destination object meets preparatory condition, to institute State face information to be identified to be labeled, face information after being marked；

Updating block, for being updated according to face information after the mark to the destination object in verification sample set.
10. device according to claim 9, it is characterised in that the recognition unit includes obtaining subelement, matching son list Member and determination subelement；

The acquisition subelement, for obtaining human face recognition model；

Coupling subelement, for by the human face recognition model, the face information to be identified to be verified into sample set with default In verification sample matched respectively, obtain the face information to be identified with it is each verification sample matching degree；

Determination subelement, for matching degree highest verification sample to be defined as to the target pair of the face information to be identified As, and the destination object and the face information to be identified and the matching degree of the destination object are tied as identification Fruit.
11. device according to claim 10, it is characterised in that the coupling subelement, be specifically used for：

By the human face recognition model, face characteristic extraction is carried out to the face information to be identified, obtains fisrt feature letter Breath；

By the human face recognition model, face characteristic extraction is carried out to the verification sample in the verification sample set, obtained each Second feature information corresponding to individual verification sample；

Fisrt feature information second feature information corresponding with each verification sample is matched respectively, obtained described to be identified Face information and the matching degree of each verification sample.
12. device according to claim 9, it is characterised in that

The mark unit, it is higher than predetermined threshold value specifically for the matching degree in the face information to be identified and destination object When, the identity of the destination object is obtained, for the identity mark of the upper destination object of face information mark to be identified Know, face information after being marked.
13. according to the device described in any one of claim 9 to 12, it is characterised in that

The updating block, specifically for calculating the equal of face information and the destination object after the mark according to preset algorithm Value, the destination object verified in sample set is replaced with the average.
14. device according to claim 10, it is characterised in that also including adding device and training unit；

The adding device, for face information after the mark to be concentrated added to labeled data；

Training unit, for according to described in labeled data set pair human face recognition model be trained, with to the recognition of face Model is updated.
15. a kind of storage medium, it is characterised in that the storage medium is stored with a plurality of instruction, and the instruction is suitable to processor Loaded, the step in the face information mask method described in 1 to 8 any one is required with perform claim.