CN109886186A - A kind of face identification method and device - Google Patents
A kind of face identification method and device Download PDFInfo
- Publication number
- CN109886186A CN109886186A CN201910123456.XA CN201910123456A CN109886186A CN 109886186 A CN109886186 A CN 109886186A CN 201910123456 A CN201910123456 A CN 201910123456A CN 109886186 A CN109886186 A CN 109886186A
- Authority
- CN
- China
- Prior art keywords
- identified
- facial image
- identification
- testimony
- witness
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The present invention provides a kind of face identification method and device, comprising: obtains facial image to be identified and certificate information to be identified；Noise suppression preprocessing is carried out to facial image to be identified, obtains pretreatment facial image, and extract the certificate image information of certificate information to be identified；Identification comparison processing is carried out to pretreatment facial image and certificate image information according to the testimony of a witness identification artificial intelligence model constructed in advance, obtains face recognition result.Face identification method and device provided by the invention can be improved the accuracy of recognition of face, and then promote recognition of face efficiency.
Description
Technical field
The present invention relates to data communication technology fields, in particular to a kind of face identification method and device.
Background technique
With the fast development of computer technology, image processing techniques, mode identification technology etc., face recognition technology is also transported
With more and more extensive.Now, the current effect in airport, railway station etc. is improved usually using the method for recognition of face verifying identity
Rate first acquires the face information and identity card picture information of the people that swipes the card, then carries out face information and identity card picture information
Matching identification.However, it has been found in practice that, often there are the feelings waited in line in the place big in flows of the people such as airport, railway stations
Condition, existing recognition of face are easy to be influenced by the external shelter such as glasses, hair and eyebrow, thereby reduce recognition of face precision.
Summary of the invention
In view of the above problems, the present invention provides a kind of face identification method and devices, can reduce the knowledge of noise on human face
Other interference improves the accuracy of recognition of face, and then promotes recognition of face efficiency.
To achieve the goals above, the present invention adopts the following technical scheme that:
First aspect present invention discloses a kind of face identification method, comprising:
Obtain facial image to be identified and certificate information to be identified；
Noise suppression preprocessing is carried out to the facial image to be identified, obtains pretreatment facial image, and extract it is described to
Identify the certificate image information of certificate information；
According to the testimony of a witness identification artificial intelligence model constructed in advance to the pretreatment facial image and the certificate image
Information carries out identification comparison processing, obtains face recognition result.
As an alternative embodiment, in first aspect present invention, it is described obtain facial image to be identified and
Before certificate information to be identified, the method also includes:
Primary loss function is constructed, and facial image and certificate are compared according to primary loss function building for identification
The original testimony of a witness identification model of image information；
Obtain the training data of the training original testimony of a witness identification model；
The original testimony of a witness identification model is trained by the training data, obtains testimony of a witness identification artificial intelligence mould
Type.
As an alternative embodiment, in first aspect present invention, by the training data to described original
Testimony of a witness identification model is trained, and obtains testimony of a witness identification artificial intelligence model, comprising:
Parameter adjustment is carried out to primary loss function according to preset adjustment rule, is adjusted rear loss function；Wherein,
The preset adjustment rule is set according to control variate method；
The training data is handled according to loss function after the adjustment, obtains loss result；
Whether loss function meets the default condition of convergence after judging the adjustment according to the loss result；
When loss function meets the default condition of convergence after the adjustment, determined according to loss function after the adjustment
The model parameter of the original testimony of a witness identification model；
Parameter adjustment is carried out to the original testimony of a witness identification model according to the model parameter, the testimony of a witness is obtained and identifies artificial intelligence
It can model.
As an alternative embodiment, in first aspect present invention, the primary loss function are as follows:
Wherein, L indicates the primary loss function, m1、m2、m3For the function parameter of the primary loss function, N is spy
The quantity of vector sample is levied, described eigenvector sample is to carry out feature extraction to the training sample to handle；
Wherein, s is predetermined coefficient, θjIndicate that j-th of feature vector sample belongs to the probability of jth class,Indicate i-th of institute
State the probability that feature vector sample belongs to the i-th class, yiIt is the true value of i-th of described eigenvector sample.
As an alternative embodiment, identifying people according to the testimony of a witness constructed in advance in first aspect present invention
Work model of mind carries out identification comparison processing to the pretreatment facial image and the certificate image information, obtains recognition of face
As a result after, the method also includes:
According to the face recognition result judge the facial image to be identified and the certificate information to be identified whether
Match；
If the facial image to be identified and the certificate information to be identified do not match that, the testimony of a witness is unmatched mentions for output
Show information.
Second aspect of the present invention discloses a kind of face identification device, comprising:
Module is obtained, for obtaining facial image to be identified and certificate information to be identified；
Preprocessing module, for obtaining pretreatment facial image to the facial image progress noise suppression preprocessing to be identified,
And extract the certificate image information of the certificate information to be identified；
Comparison module is identified, for identifying artificial intelligence model to the pretreatment face figure according to the testimony of a witness constructed in advance
Picture and the certificate image information carry out identification comparison processing, obtain face recognition result.
As an alternative embodiment, in second aspect of the present invention, the face identification device further include:
Module is constructed, for constructing original damage before acquisition facial image to be identified and certificate information to be identified
Function is lost, and compares the primitive man of facial image and certificate image information for identification according to primary loss function building
Demonstrate,prove identification model；
The acquisition module is also used to obtain the training data of the training original testimony of a witness identification model；
Training module obtains the testimony of a witness for being trained by the training data to the original testimony of a witness identification model
Identify artificial intelligence model.
As an alternative embodiment, in second aspect of the present invention, the face identification device further include:
Judgment module, for identifying artificial intelligence model to the pretreatment facial image according to the testimony of a witness constructed in advance
Identification comparison processing is carried out with the certificate image information, after obtaining face recognition result, according to the face recognition result
Judge whether the facial image to be identified and the certificate information to be identified match；
Cue module, for when the facial image to be identified and the certificate information to be identified do not match that, output
The unmatched prompt information of the testimony of a witness.
Third aspect present invention discloses a kind of computer equipment, including memory and processor, and the memory is used for
Computer program is stored, the processor runs the computer program so that the computer equipment executes first aspect and discloses
The some or all of face identification method.
Fourth aspect present invention discloses a kind of computer readable storage medium, is stored with computer described in the third aspect
The computer program used in equipment.
The face identification method and device provided according to the present invention is obtaining facial image to be identified and certificate to be identified letter
After breath, noise suppression preprocessing first is carried out to facial image to be identified and is obtained with removing the noise jamming in facial image to be identified
Facial image is pre-processed, and extracts the certificate image information of certificate information to be identified；Finally, further according to the testimony of a witness constructed in advance
Identification artificial intelligence model carries out identification comparison processing to pretreatment facial image and certificate image information, obtains recognition of face knot
Fruit can reduce the interference that noise on human face identifies in facial image to be identified, improve the accuracy of recognition of face, and then promoted
Recognition of face efficiency.
To enable the above objects, features and advantages of the present invention to be clearer and more comprehensible, preferred embodiment is cited below particularly, and cooperate
Appended attached drawing, is described in detail below.
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 understood that the following drawings illustrates only certain embodiments of the present invention, therefore is not construed as pair
The restriction of the scope of the invention.
Fig. 1 is a kind of flow diagram for face identification method that the embodiment of the present invention one provides；
Fig. 2 is a kind of flow diagram of face identification method provided by Embodiment 2 of the present invention；
Fig. 3 is a kind of structural schematic diagram for face identification device that the embodiment of the present invention three provides.
Specific embodiment
Below in conjunction with attached drawing in the embodiment of the present invention, technical solution in the embodiment of the present invention carries out clear, complete
Ground description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Usually exist
The component of the embodiment of the present invention described and illustrated in attached drawing can be arranged and be designed with a variety of different configurations herein.Cause
This, is not intended to limit claimed invention to the detailed description of the embodiment of the present invention provided in the accompanying drawings below
Range, but it is merely representative of selected embodiment of the invention.Based on the embodiment of the present invention, those skilled in the art are not doing
Every other embodiment obtained under the premise of creative work out, shall fall within the protection scope of the present invention.
For the problems of the prior art, the present invention provides a kind of face identification method and devices；It is to be identified obtaining
After facial image and certificate information to be identified, noise suppression preprocessing first is carried out to facial image to be identified, to remove people to be identified
Noise jamming in face image obtains pretreatment facial image, and extracts the certificate image information of certificate information to be identified；Most
Afterwards, pretreatment facial image and certificate image information are identified further according to the testimony of a witness identification artificial intelligence model constructed in advance
Comparison processing, obtains face recognition result, can reduce the interference that noise on human face identifies in facial image to be identified, improves people
The accuracy of face identification, and then promote recognition of face efficiency.Also, the technology can use relevant software or hardware realization,
It is described below by embodiment.
Embodiment 1
Referring to Fig. 1, Fig. 1 is a kind of flow diagram of face identification method provided in an embodiment of the present invention.Wherein, such as
Shown in Fig. 1, which be may comprise steps of:
S101, facial image to be identified and certificate information to be identified are obtained.
In the present embodiment, which can be applied to the public places such as station, railway station, school.Required prison
The public place of control is equipped with photographic device and reader device, and facial image to be identified can be obtained by photographic device, is passed through
Reader device obtains certificate information to be identified.
S102, noise suppression preprocessing is carried out to facial image to be identified, obtains pretreatment facial image, and extract to be identified
The certificate image information of certificate information.
In the embodiment of the present application, after obtaining facial image to be identified and certificate information to be identified, need to be identified
Facial image, which carries out noise suppression preprocessing, can be avoided due to blocking and produces to remove the noise jamming in facial image to be identified
The problems such as raw noise jamming, and then promote human body behavior prediction precision.
As an alternative embodiment, carrying out noise suppression preprocessing to facial image to be identified, light benefit is specifically included
It repays, greyscale transformation, histogram equalization, normalization, geometric correction, filtering processing, Edge contrast, the processing such as size scaling, it is right
This, the present embodiment is not limited in any way.
In the above-described embodiment, image gray-scale transformation can be used by carrying out greyscale transform process to facial image to be identified
Method can remove the noise jamming in facial image to be identified, and then improve the image quality of facial image to be identified, make to be identified
The image display effect of facial image is more clear, to promote the accuracy of subsequent behavior prediction.
In the above-described embodiment, picture size scaling can be carried out to facial image to be identified according to preset image sizes
Processing, so as to the different original video size of compatibility.
In the above-described embodiment, bilinear interpolation can be used by carrying out picture size scaling processing to greyscale transformation image
Algorithm, bicubic interpolation algorithm etc., are not limited in any way this present embodiment.
In the above-described embodiment, face cutting process can also be carried out to facial image to be identified, it is to be identified to extract
Facial image in facial image.
S103, identify artificial intelligence model to pretreatment facial image and certificate image information according to the testimony of a witness constructed in advance
Identification comparison processing is carried out, face recognition result is obtained.
In the face identification method described in Fig. 1, after obtaining facial image to be identified and certificate information to be identified,
First noise suppression preprocessing is carried out to facial image to be identified to be pre-processed to remove the noise jamming in facial image to be identified
Facial image, and extract the certificate image information of certificate information to be identified；Finally, identifying people further according to the testimony of a witness constructed in advance
Work model of mind carries out identification comparison processing to pretreatment facial image and certificate image information, obtains face recognition result.It can
See, implement face identification method described in Fig. 1, the interference that noise on human face identifies in facial image to be identified can be reduced,
The accuracy of recognition of face is improved, and then promotes recognition of face efficiency.
Embodiment 2
Referring to Fig. 2, Fig. 2 is a kind of flow diagram of face identification method provided in an embodiment of the present invention.Wherein, such as
Shown in Fig. 2, which be may comprise steps of:
S201, building primary loss function, and facial image and card are compared according to the building of primary loss function for identification
The original testimony of a witness identification model of part image information.
As an alternative embodiment, primary loss function are as follows:
Wherein, L indicates primary loss function, m1、m2、m3For the function parameter of primary loss function, N is feature vector sample
This quantity, feature vector sample are to carry out feature extraction to training sample to handle；
Wherein, s is predetermined coefficient, θjIndicate that j-th of feature vector sample belongs to the probability of jth class,Indicate i-th of spy
Sign vector sample belongs to the probability of the i-th class, yiIt is the true value of ith feature vector sample.
In the above-described embodiment, m1∈ [1,5], m2∈ [0,0.5], m3∈ [0,0.4], does not appoint this present embodiment
What is limited.
In the above-described embodiment, above-mentioned primary loss function is losses by mixture function, is SphereFace,
The combination of CosinFace and Arcface.Although parameter (the i.e. m to be adjusted of the losses by mixture function1、m2And m3) more, it is single
It is relatively difficult that private losses by mixture function training network may also will lead to tune ginseng, it is not easy to it restrains, but in SphereFace,
On the basis of CosinFace and Arcface training, then m is determined respectively1、m2、m3Suitable value range and efficient combination,
Often there is good effect for promoting testimony of a witness matching identification accuracy rate, and for different data sets, adaptability
It is stronger.
In the above-described embodiment, predetermined coefficient s is artificially to preset, and specifically can be set to 64, compares this implementation
Example is not limited in any way.
In the above-described embodiment, θjIndicate that j-th of feature vector sample belongs to the probability of jth class, formula are as follows:
Wherein, W indicates the weight of j-th of classification.
In the above-described embodiment, original testimony of a witness identification model includes that primitive character extracts network and sorter network.If X0,
X1, X2..., Xi..., XNFor N number of training sample, then network is extracted by primitive character and feature extraction is carried out to N number of training sample
Processing, available N number of feature vector sample, i.e. x0, x1, x2..., xi..., xN。
S202, the training data for obtaining the original testimony of a witness identification model of training.
S203, original testimony of a witness identification model is trained by training data, obtains testimony of a witness identification artificial intelligence model.
As an alternative embodiment, being trained by training data to original testimony of a witness identification model, people is obtained
Card identification artificial intelligence model, comprising:
Parameter adjustment is carried out to primary loss function according to preset adjustment rule, is adjusted rear loss function；Wherein,
Preset adjustment rule is set according to control variate method；
Training data is handled according to loss function after adjustment, obtains loss result；
Whether loss function meets the default condition of convergence after judging adjustment according to loss result；
When loss function meets the default condition of convergence after adjustment, determine that the original testimony of a witness identifies according to loss function after adjustment
The model parameter of model；
Parameter adjustment is carried out to original testimony of a witness identification model according to model parameter, obtains testimony of a witness identification artificial intelligence model.
In the above-described embodiment, the m of primary loss function can be determined according to preset adjustment rule1、m2And m3Three
Parameter, then by m1、m2And m3Three parameters substitute into above-mentioned primary loss function formula, are adjusted rear loss function.
As an alternative embodiment, m1、m2And m3The value range of three parameters can be with are as follows: m1∈ [1,5], m2∈
[0,0.5], m3∈ [0,0.4], is not limited in any way this present embodiment.
In the above-described embodiment, parameter adjustment is carried out to primary loss function according to preset adjustment rule, is adjusted
Loss function after whole.Specific adjustment process is to obtain preset original value (such as m first1=5, m2=0.5, m3=0.4),
Then it is taken turns cycle of training by one, fixed m1、m2And m3Two of them parameter is (such as fixed m1=5, m2=0.5) value, then again
Another parameter is adjusted (such as fixed m according to preset adjustment law1=5, m2After=0.5, m is adjusted3) value, and so on,
It can finally determine the m of primary loss function1、m2And m3Three parameters.
In the above-described embodiment, it is assumed that fixed m1=5, m2After=0.5, m is adjusted3Value, then preset adjustment law
It can be m3It is adjusted with reducing 0.1 every wheel cycle of training；Assuming that fixed m2=0.5, m3After=0.4, m is adjusted1Value,
Then preset adjustment law can be m1It is adjusted with reducing 1 every wheel cycle of training；Assuming that fixed m1=5, m3After=0.4,
Adjust m2Value, then preset adjustment law can be m2It is adjusted with reducing 0.1 every wheel cycle of training, to this present embodiment
It is not limited in any way.
Preferably, predetermined coefficient s=64, m1=1, m2=0.3, m3=0.2.
S204, facial image to be identified and certificate information to be identified are obtained.
S205, noise suppression preprocessing is carried out to facial image to be identified, obtains pretreatment facial image, and extract to be identified
The certificate image information of certificate information.
S206, identify artificial intelligence model to pretreatment facial image and certificate image information according to the testimony of a witness constructed in advance
Identification comparison processing is carried out, face recognition result is obtained.
S207, judge whether facial image to be identified and certificate information to be identified match according to face recognition result, if
It is to terminate this process；If mismatched, step S208 is executed.
S208, the output unmatched prompt information of the testimony of a witness.
As it can be seen that implementing face identification method described in Fig. 2, noise on human face in facial image to be identified can be reduced and known
Other interference improves the accuracy of recognition of face, and then promotes recognition of face efficiency；On the other hand, do not identifying the testimony of a witness not
Timing can export warning message in time.
Embodiment 3
Referring to Fig. 3, Fig. 3 is a kind of structural schematic diagram of face identification device provided in an embodiment of the present invention.Wherein, such as
Shown in Fig. 3, which includes:
Module 301 is obtained, for obtaining facial image to be identified and certificate information to be identified.
Preprocessing module 302, for obtaining pretreatment facial image to facial image to be identified progress noise suppression preprocessing,
And extract the certificate image information of certificate information to be identified.
Comparison module 303 is identified, for identifying artificial intelligence model to pretreatment face figure according to the testimony of a witness constructed in advance
Picture and certificate image information carry out identification comparison processing, obtain face recognition result.
As an alternative embodiment, the face identification device further include:
Module 304 is constructed, for constructing primary loss before obtaining facial image to be identified and certificate information to be identified
Function, and identified according to the original testimony of a witness that the building of primary loss function compares facial image and certificate image information for identification
Model.
Module 301 is obtained, is also used to obtain the training number of the original testimony of a witness identification model of training.
Training module 305 obtains testimony of a witness identification people for being trained by training data to original testimony of a witness identification model
Work model of mind.
In the present embodiment, training module 305 is obtaining testimony of a witness identification artificial intelligence model, can also trigger acquisition module
301 obtain facial image to be identified and certificate information to be identified.
As an alternative embodiment, primary loss function are as follows:
Wherein, L indicates primary loss function, m1、m2、m3For the function parameter of primary loss function, N is feature vector sample
This quantity, feature vector sample are to carry out feature extraction to training sample to handle；
Wherein, s is predetermined coefficient, θjIndicate that j-th of feature vector sample belongs to the probability of jth class,Indicate i-th of spy
Sign vector sample belongs to the probability of the i-th class, yiIt is the true value of ith feature vector sample.
In the above-described embodiment, m1∈ [1,5], m2∈ [0,0.5], m3∈ [0,0.4], does not appoint this present embodiment
What is limited.
Preferably, predetermined coefficient s=64, m1=1, m2=0.3, m3=0.2.
As an alternative embodiment, the face identification device further include:
Judgment module 306, for identifying artificial intelligence model pair according to the testimony of a witness constructed in advance in identification comparison module 303
Pretreatment facial image and certificate image information carry out identification comparison processing, after obtaining face recognition result, are known according to face
Other result judges whether facial image to be identified and certificate information to be identified match.
Cue module 307, for judging facial image to be identified and certificate information to be identified not phase when judgment module 306
When matching, the unmatched prompt information of the testimony of a witness is exported.
As it can be seen that face identification device described in implementing Fig. 3, can reduce noise on human face in facial image to be identified and know
Other interference improves the accuracy of recognition of face, and then promotes recognition of face efficiency.
In addition, the present invention also provides a kind of computer equipments.The computer equipment includes memory and processor, storage
Device can be used for storing computer program, and processor is by operation computer program, so that the computer equipment be made to execute above-mentioned side
The function of method or the modules in above-mentioned face identification device.
Memory may include storing program area and storage data area, wherein storing program area can storage program area, at least
Application program needed for one function (such as sound-playing function, image player function etc.) etc.；Storage data area can store root
Created data (such as audio data, phone directory etc.) etc. are used according to mobile terminal.In addition, memory may include high speed
Random access memory, can also include nonvolatile memory, a for example, at least disk memory, flush memory device or
Other volatile solid-state parts.
The present embodiment additionally provides a kind of computer storage medium, for storing calculating used in above-mentioned computer equipment
Machine program.
In several embodiments provided herein, it should be understood that disclosed device and method can also pass through
Other modes are realized.The apparatus embodiments described above are merely exemplary, for example, flow chart and structure in attached drawing
Figure shows the system frame in the cards of the device of multiple embodiments according to the present invention, method and computer program product
Structure, function and operation.In this regard, each box in flowchart or block diagram can represent a module, section or code
A part, a part of the module, section or code includes one or more for implementing the specified logical function
Executable instruction.It should also be noted that function marked in the box can also be to be different from the implementation as replacement
The sequence marked in attached drawing occurs.For example, two continuous boxes can actually be basically executed in parallel, they are sometimes
It can execute in the opposite order, this depends on the function involved.It is also noted that in structure chart and/or flow chart
The combination of each box and the box in structure chart and/or flow chart, can function or movement as defined in executing it is dedicated
Hardware based system realize, or can realize using a combination of dedicated hardware and computer instructions.
In addition, each functional module or unit in each embodiment of the present invention can integrate one independence of formation together
Part, be also possible to modules individualism, an independent part can also be integrated to form with two or more modules.
It, can be with if the function is realized and when sold or used as an independent product in the form of software function module
It is stored in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially in other words
The part of the part that contributes to existing technology or the technical solution can be embodied in the form of software products, the meter
Calculation machine software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be intelligence
Can mobile phone, personal computer, server or simulator etc.) execute each embodiment the method for the present invention whole or
Part steps.And storage medium above-mentioned include: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory),
Random access memory (RAM, Random Access Memory), magnetic or disk etc. be various to can store program code
Medium.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. a kind of face identification method characterized by comprising
Obtain facial image to be identified and certificate information to be identified；
Noise suppression preprocessing is carried out to the facial image to be identified, obtains pretreatment facial image, and extract described to be identified
The certificate image information of certificate information；
According to the testimony of a witness identification artificial intelligence model constructed in advance to the pretreatment facial image and the certificate image information
Identification comparison processing is carried out, face recognition result is obtained.
2. face identification method according to claim 1, which is characterized in that it is described obtain facial image to be identified and to
Before identifying certificate information, the method also includes:
Primary loss function is constructed, and facial image and certificate image are compared according to primary loss function building for identification
The original testimony of a witness identification model of information；
Obtain the training data of the training original testimony of a witness identification model；
The original testimony of a witness identification model is trained by the training data, obtains testimony of a witness identification artificial intelligence model.
3. face identification method according to claim 2, which is characterized in that by the training data to the primitive man
Card identification model is trained, and obtains testimony of a witness identification artificial intelligence model, comprising:
Parameter adjustment is carried out to primary loss function according to preset adjustment rule, is adjusted rear loss function；Wherein, described
Preset adjustment rule is set according to control variate method；
The training data is handled according to loss function after the adjustment, obtains loss result；
Whether loss function meets the default condition of convergence after judging the adjustment according to the loss result；
When loss function meets the default condition of convergence after the adjustment, according to loss function determination after the adjustment
The model parameter of original testimony of a witness identification model；
Parameter adjustment is carried out to the original testimony of a witness identification model according to the model parameter, obtains testimony of a witness identification artificial intelligence mould
Type.
4. face identification method according to claim 2, which is characterized in that the primary loss function are as follows:
Wherein, L indicates the primary loss function, m1、m2、m3For the function parameter of the primary loss function, N be characterized to
The quantity of sample is measured, described eigenvector sample is to carry out feature extraction to the training sample to handle；
Wherein, s is predetermined coefficient, θjIndicate that j-th of feature vector sample belongs to the probability of jth class, θyiIndicate i-th of spy
Sign vector sample belongs to the probability of the i-th class, yiIt is the true value of i-th of described eigenvector sample.
5. face identification method according to claim 1, which is characterized in that artificial according to the testimony of a witness identification constructed in advance
Model of mind carries out identification comparison processing to the pretreatment facial image and the certificate image information, obtains recognition of face knot
After fruit, the method also includes:
Judge whether the facial image to be identified and the certificate information to be identified match according to the face recognition result；
If the facial image to be identified and the certificate information to be identified do not match that, the unmatched prompt letter of the output testimony of a witness
Breath.
6. a kind of face identification device characterized by comprising
Module is obtained, for obtaining facial image to be identified and certificate information to be identified；
Preprocessing module is used to carry out noise suppression preprocessing to the facial image to be identified, obtains pretreatment facial image, and
Extract the certificate image information of the certificate information to be identified；
Identify comparison module, for according to construct in advance the testimony of a witness identification artificial intelligence model to the pretreatment facial image with
The certificate image information carries out identification comparison processing, obtains face recognition result.
7. face identification device according to claim 6, which is characterized in that further include:
Module is constructed, for constructing primary loss letter before acquisition facial image to be identified and certificate information to be identified
Number, and known according to the original testimony of a witness that primary loss function building compares facial image and certificate image information for identification
Other model；
The acquisition module is also used to obtain the training data of the training original testimony of a witness identification model；
Training module obtains testimony of a witness identification for being trained by the training data to the original testimony of a witness identification model
Artificial intelligence model.
8. face identification device according to claim 6, which is characterized in that further include:
Judgment module, for identifying artificial intelligence model to the pretreatment facial image and institute according to the testimony of a witness constructed in advance
It states certificate image information and carries out identification comparison processing, after obtaining face recognition result, judged according to the face recognition result
Whether the facial image to be identified and the certificate information to be identified match；
Cue module, for exporting the testimony of a witness when the facial image to be identified and the certificate information to be identified do not match that
Unmatched prompt information.
9. a kind of computer equipment, which is characterized in that including memory and processor, the memory is for storing computer
Program, the processor runs the computer program so that the computer equipment perform claim requires any one of 1 to 5 institute
The face identification method stated.
10. a kind of computer readable storage medium, which is characterized in that it is stored in computer equipment as claimed in claim 9
The used computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910123456.XA CN109886186A (en) | 2019-02-18 | 2019-02-18 | A kind of face identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910123456.XA CN109886186A (en) | 2019-02-18 | 2019-02-18 | A kind of face identification method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109886186A true CN109886186A (en) | 2019-06-14 |
Family
ID=66928456
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910123456.XA Pending CN109886186A (en) | 2019-02-18 | 2019-02-18 | A kind of face identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109886186A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111310509A (en) * | 2020-03-12 | 2020-06-19 | 北京大学 | Real-time bar code detection system and method based on logistics waybill |
| CN111626091A (en) * | 2020-03-09 | 2020-09-04 | 咪咕文化科技有限公司 | Face image annotation method and device and computer readable storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106339673A (en) * | 2016-08-19 | 2017-01-18 | 中山大学 | ATM identity authentication method based on face recognition |
| CN106951867A (en) * | 2017-03-22 | 2017-07-14 | 成都擎天树科技有限公司 | Face identification method, device, system and equipment based on convolutional neural networks |
| CN108009528A (en) * | 2017-12-26 | 2018-05-08 | 广州广电运通金融电子股份有限公司 | Face authentication method, device, computer equipment and storage medium based on Triplet Loss |
| CN108875559A (en) * | 2018-04-27 | 2018-11-23 | 中国科学院自动化研究所 | The face identification method and system shone based on certificate photo and scene |
-
2019
- 2019-02-18 CN CN201910123456.XA patent/CN109886186A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106339673A (en) * | 2016-08-19 | 2017-01-18 | 中山大学 | ATM identity authentication method based on face recognition |
| CN106951867A (en) * | 2017-03-22 | 2017-07-14 | 成都擎天树科技有限公司 | Face identification method, device, system and equipment based on convolutional neural networks |
| CN108009528A (en) * | 2017-12-26 | 2018-05-08 | 广州广电运通金融电子股份有限公司 | Face authentication method, device, computer equipment and storage medium based on Triplet Loss |
| CN108875559A (en) * | 2018-04-27 | 2018-11-23 | 中国科学院自动化研究所 | The face identification method and system shone based on certificate photo and scene |
Non-Patent Citations (4)
| Title |
|---|
| JIANKANG DENG ET AL: "ArcFace: Additive Angular Margin Loss for Deep Face Recognition", 《ARXIV:1801.07698V3》 * |
| NTTSTAR: "arcface", 《HTTPS://GITHUB.COM/DEEPINSIGHT/INSIGHTFACE/TREE/MASTER/RESOURCES》 * |
| 蔡晨伟: "基于深度特征学习与分类器关联集成的人证比对系统研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 都伊林: "《智能安防新发展与应用》", 31 July 2018 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111626091A (en) * | 2020-03-09 | 2020-09-04 | 咪咕文化科技有限公司 | Face image annotation method and device and computer readable storage medium |
| CN111626091B (en) * | 2020-03-09 | 2023-09-22 | 咪咕文化科技有限公司 | Face image labeling method and device and computer readable storage medium |
| CN111310509A (en) * | 2020-03-12 | 2020-06-19 | 北京大学 | Real-time bar code detection system and method based on logistics waybill |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109697416B (en) | Video data processing method and related device | |
| JP7490141B2 (en) | IMAGE DETECTION METHOD, MODEL TRAINING METHOD, IMAGE DETECTION APPARATUS, TRAINING APPARATUS, DEVICE, AND PROGRAM | |
| CN103473564B (en) | A kind of obverse face detection method based on sensitizing range | |
| CN107333071A (en) | Video processing method and device, electronic equipment and storage medium | |
| WO2016138838A1 (en) | Method and device for recognizing lip-reading based on projection extreme learning machine | |
| CN107578380A (en) | Image processing method and device, electronic equipment and storage medium | |
| CN111310718A (en) | High-accuracy detection and comparison method for face-shielding image | |
| CN113343826A (en) | Training method of human face living body detection model, human face living body detection method and device | |
| CN111951154B (en) | Picture generation method and device containing background and medium | |
| US20180137630A1 (en) | Image processing apparatus and method | |
| CN105528757A (en) | Content-based image aesthetic quality improvement method | |
| CN111160110A (en) | Method and device for identifying anchor based on face features and voice print features | |
| CN110648289A (en) | Image denoising processing method and device | |
| CN110210278A (en) | A kind of video object detection method, device and storage medium | |
| CN109145720A (en) | A kind of face identification method and device | |
| CN114913588A (en) | Face image restoration and recognition method applied to complex scene | |
| CN108197669A (en) | The feature training method and device of convolutional neural networks | |
| CN114596608A (en) | Double-stream video face counterfeiting detection method and system based on multiple clues | |
| CN111353364A (en) | Dynamic face identification method and device and electronic equipment | |
| CN109886186A (en) | A kind of face identification method and device | |
| CN117975519A (en) | Model training and image generating method and device, electronic equipment and storage medium | |
| CN116012908A (en) | Face generation method and device, electronic equipment and storage medium | |
| CN110688972B (en) | System and method for improving face generation performance | |
| CN109145737B (en) | Rapid face recognition method and device, electronic equipment and storage medium | |
| Belahcene et al. | 3D and 2D face recognition based on image segmentation |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190614 |