CN106096538A - Face identification method based on sequencing neural network model and device - Google Patents
Face identification method based on sequencing neural network model and device Download PDFInfo
- Publication number
- CN106096538A CN106096538A CN201610403028.9A CN201610403028A CN106096538A CN 106096538 A CN106096538 A CN 106096538A CN 201610403028 A CN201610403028 A CN 201610403028A CN 106096538 A CN106096538 A CN 106096538A
- Authority
- CN
- China
- Prior art keywords
- face
- sequencing
- image
- neural network
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000012163 sequencing technique Methods 0.000 title claims abstract description 54
- 238000003062 neural network model Methods 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000012549 training Methods 0.000 claims abstract description 49
- 230000001815 facial effect Effects 0.000 claims abstract description 43
- 230000007935 neutral effect Effects 0.000 claims abstract description 21
- 238000012937 correction Methods 0.000 claims abstract description 5
- 238000013527 convolutional neural network Methods 0.000 claims description 10
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 238000005286 illumination Methods 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 2
- 238000013528 artificial neural network Methods 0.000 abstract description 3
- 239000000523 sample Substances 0.000 description 25
- 230000006870 function Effects 0.000 description 8
- 230000000694 effects Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 230000001537 neural effect Effects 0.000 description 5
- 230000004913 activation Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000012497 inhomogeneous sample Substances 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
- G06V40/165—Detection; Localisation; Normalisation using facial parts and geometric relationships
Abstract
The present invention discloses a kind of face identification method based on sequencing neural network model and device.The method includes: the facial image of input is carried out pretreatment operation, the angle of correction facial image and expression；The neutral net comprising ordering operation is used to extract the feature having corrected facial image/video；Feature representation according to facial image calculates the similarity between image pair, thus learns the identity of special object in input facial image.The present invention is directed in recognition of face problem, human face recognition model parameter based on neutral net is many, the problem that computing cost is big, proposes sequencing neural network structure, is represented by the sequencing between different characteristic and efficiently reduce network parameter, saves the calculating time；And for the less problem of training data, it is proposed that based on contrast loss, the training method of tlv triple loss.
Description
Technical field
The present invention relates to the technical fields such as artificial intelligence, pattern recognition, Digital Image Processing, be specifically related to a kind of based on fixed
The face identification method of sequence neural network model and device.
Background technology
As the one of biometrics identification technology, recognition of face owing to it is untouchable and gathers convenient feature,
There is good development and application prospect.Face recognition technology has all played highly important effect in many application scenarios,
Such as airport security, frontier inspection clearance etc..In recent years along with the high speed development of the Internet finance, face recognition technology is in mobile payment
On show great application advantage.The purpose of recognition of face is to learn user's according to the user's facial image obtained or video
Identity.At present, face recognition technology still cannot meet real requirement under outdoor uncontrolled environment, and its Major Difficulties is illumination
Change, user's attitude expression shape change, age pattern of body form change and block.
In recent years, degree of depth study all achieves, in the various fields of machine vision, the effect attracted people's attention.Look steadily the most
Purpose model surely belongs to convolutional neural networks, and this class model uses multilamellar convolutional layer and pond layer, can be with abstract image or video counts
Effective hierarchical feature according to, it is achieved stronger non-linear expression.Convolutional neural networks is at object classification, action recognition, figure
As fields such as segmentation and recognitions of face, all achieve the effect being significantly stronger than traditional method.In some Low Level Vision problems,
Such as image denoising, in the problems such as image super-resolution strengthens, image deblurring, degree of depth learning art the most all achieves good
Effect.In field of face identification, face identification method based on neutral net and degree of depth study also due to the performance of its excellence and
Receiving much concern, front-runner's face recognizer is mostly based on degree of depth learning model the most both at home and abroad.Face based on degree of depth study
Recognition methods is generally divided into two steps: first by neural network model, the facial image of input is calculated a mark sheet
Reach；Then facial image is obtained according to the similarity between feature representation.
Along with the arriving of big data age, the data scale that we need to process is the biggest, the speed of face recognition algorithms
The high efficiency ever more important of degree.Especially in mobile payment field, request memory and the speed of face recognition algorithms directly affect
The waiting time of user.Therefore, at present in the urgent need to a kind of face recognition algorithms of exploitation, it can ensure the same of high discrimination
Time, meet lightweight, requirement that Generalization Capability is high.
Summary of the invention
(1) to solve the technical problem that
In order to solve to improve the accuracy rate of face recognition algorithms, ensureing recognizer rapidly and efficiently, the present invention carries simultaneously
Go out a kind of face identification method based on sequencing neural network model.Use sequencing neural unit, by keeping different level
Ordering relationship between feature excavates the validity feature in input picture or video.The spy self possessed due to sequencing neural unit
Levy selectivity characteristic so that the neural network model often parameter amount comprising sequencing neural unit is less, possesses the feature of lightweight,
Thus ensure that face recognition algorithms calculates speed and less storage demand faster.
(2) technical scheme
The present invention proposes a kind of face identification method based on sequencing neural network model, including:
Step S1, the image to be identified of reading input, detect the face location in image to be identified and key point confidence
Breath；
Step S2, according to described face location information and key point information, image to be identified is carried out pretreatment operation；
Step S3, by the input of pretreated image to be identified in sequencing neural network model, obtain image to be identified
Feature representation；
Step S4, the feature representation calculating image to be identified and the similarity of known facial image feature in data base, with
Identify image to be identified.
The invention allows for a kind of face identification device based on sequencing neural network model, it is characterised in that including:
Input module, for reading in the image to be identified of input, detects the face location in image to be identified and key point
Positional information；
Pretreatment module, for carrying out pretreatment according to described face location information and key point information to image to be identified
Operation；
Feature acquisition module, for by pretreated image to be identified input to sequencing neural network model, obtaining
The feature representation of image to be identified；
Identification module, for calculating the similar of the feature representation facial image feature known to data base of image to be identified
Degree, to identify image to be identified.
The present invention is directed in recognition of face problem, human face recognition model parameter based on neutral net is many, and computing cost is big
Problem, propose sequencing neural network structure, represented by the sequencing between different characteristic and efficiently reduce network parameter, save meter
Evaluation time；And for the less problem of training data, it is proposed that based on contrast loss, the training method of tlv triple loss.This
The sequencing neural network model of bright employing can be used in the problems such as image/video classification, image retrieval, recognition of face, is protecting
While card high-accuracy, that compares existing neural network model possesses less network parameter so that carrying cost, calculating
Cost is substantially reduced, and is adaptive to each task under big data scene.For recognition of face, the method that invention is used is not
Only effectively reduce the parameter amount of neural network model, and the generalization ability of face representation can be obviously improved, compare equivalent parameters
Amount is greatly improved face recognition accuracy rate for the model of the time of calculating.
Accompanying drawing explanation
Fig. 1 is the method flow diagram of face identification method based on sequencing neural network model in the present invention.
Detailed description of the invention
For making the object, technical solutions and advantages of the present invention clearer, below in conjunction with instantiation, and with reference in detail
Thin accompanying drawing, the present invention is described in more detail.But described examples of implementation are intended merely to facilitate the understanding of the present invention, and right
It does not play any restriction effect.
Fig. 1 is the method flow diagram of the face identification method based on sequencing neural network model that the present invention proposes, such as Fig. 1
Shown in, the face identification method based on sequencing neural network model that the present invention proposes includes following step:
Step S1, the facial image reading in input or video, the face location information in detection input picture or frame of video
With key point positional information；
In one embodiment, according to facial image or the video of described input, application image recognizer detection face position
Confidence ceases, and according to gained face positional information, application image recognizer obtains the key point positional information of face.Wherein,
The key point of face is predefined, such as eyes, nose, mouth profile, face week profile etc..
Step S2, according to described face location information and key point positional information to the face in input picture or frame of video
Image carries out pretreatment operation.Described pretreatment operation includes attitude updating and light correction；
In one embodiment, described attitude updating comprises determining that key point position and the illumination condition of standard face, then root
The face key point position of input picture is aligned to standard face by the face location information and the key point information that obtain according to step S1
Key point position, to reach to correct the purpose of human face posture；Wherein, key point position and the illumination bar of standard face can be pre-defined
Part, or be directly used in training set and be calculated average face as standard face, it is then determined that its key point positional information and light
According to condition standard face；
In one embodiment, the correction of described light includes by image processing algorithm, by the light change of facial image extremely
Consistent with standard face.
The number of operations that attitude updating corrects with light does not limits, and its sequencing is interchangeable.
Step S3, facial image will be obtained through the input of pretreated facial image in sequencing neural network model
Feature representation；Described sequencing neural network model first passes through training in advance and obtains.
Further, described step S3 includes:
Step S3-1, training one are used for from the sequencing calculating feature representation through the facial image of pretreatment neural
Network model.Described neural network model comprises sequencing neutral net unit.Sequencing neutral net unit is neural network model
In a Class Activation function.The activation primitive of the single-input single-outputs such as sigmoi, ReLU conventional in learning from the degree of depth is different, fixed
Sequence neutral net unit takes the form of multi input, can obtain the sequencing between multiple input and express.
In one embodiment, a canonic form of described sequencing neutral net unit is:
Wherein I1, I2Being respectively two inputs of sequencing neutral net unit, Y is the output of sequencing neutral net unit.I,
J is input, output image index on two x, y direction respectively.Can be seen that from above formula, the maximum of sequencing neutral net unit
Value Operations is that step-by-step is carried out, and each is each maximum inputting corresponding position i.e. to export image.This sequencing neutral net list i.e.
Unit is output as expressing a sequencing of input.It should be noted that and formula takes maxima operation not sequencing nerve list
The sole operation form of unit, maxima operation also can be replaced the common mathematical operations such as minima, meansigma methods and poor, long-pending business.Fixed
The input of sequence neutral net unit is also not necessarily limited to two, can be the combination of multiple input.I.e. sequencing neutral net unit also may be used
Expand to following form:
Sequencing neutral net uses sequencing neutral net unit as activation primitive, can automatically learn output and appoint target
The sequencing being engaged between effective feature is expressed.
Concrete, step S3-1 farther includes:
Softmax loss, contrast loss and tlv triple loss is used to train a face characteristics of image as object function
Extraction model, the facial image x that input is normal size of this network, it is output as the facial image feature representation f of regular length
(x)。
The training data that step S3-1-1, utilization are collected, training convolutional neural networks disaggregated model, it is used for training sample
In facial image classification.The output layer of convolutional neural networks disaggregated model is a grader, and it is special that it accepts facial image
Levying expression f (x) conduct input, output valve can be used for the classification of calculating input image.For having the disaggregated model of N class, network
Output there is N number of node.
In this step, utilize a series of Classification Loss functions that softmax loss function is representative as optimization aim, instruction
Getting facial image disaggregated model, wherein softmax loss function is as follows:
Wherein, N is class number；X is input facial image；y∈RN×1It is the categorization vector representing facial image classification,
If training sample belongs to the i-th class, then only having i-th dimension in categorization vector y is 1, and other dimensions are equal to 0；Represent Neural Network Science
The grader arrived,Represent the neutral net output at output layer i-th node.
S3-1-2: using the facial image Classification Neural model that trained in step S3-1-1 as the mould of pre-training
Type, uses contrast loss (contrastive loss) and tlv triple loss (triplet loss) to continue neural network model
It is optimized training.
Removing training in S3-1-1 and obtain the output layer of model, output f (x) obtaining model rest layers is facial image
Feature representation.
The optimization aim of contrast loss is:
The input of disaggregated model output layer during wherein f (x) is step S3-1-1, the output of the most secondary last layer, for facial image
Feature representation.D (.) can be with L2 distance, the COS distance a series of distance functions as representative.During training, can be random
The sample that combined training is concentrated, sample pair between sample pair and class in structure class.θ is to control a ginseng of between class distance difference in class
Number, is set based on experience value.
The optimization aim of tlv triple loss is:
L=max (d (f (xa), f (xp))-d(f(xa)+α, f (xn)), 0)
The input of disaggregated model output layer during wherein f (x) is step S3-1-1, the output of the most secondary last layer, for facial image
Feature representation.D (.) can be with L2 distance, the COS distance a series of distance functions as representative.xaIt is referred to as central sample, xpFor
With central sample xaBelong to of a sort positive sample, xnFor with central sample xaBelong to inhomogeneous negative sample.A is for controlling in class
One parameter of between class distance difference, is set based on experience value.During training, can training set in random choose sample
This is as central sample, then chooses similar with it/inhomogeneous sample in residue sample and builds tlv triple.
In one embodiment, during the training of step S3-1-2, filter out difficult sample and combine as training data.Difficult sample
Concretely comprising the following steps of this screening:
For Large Copacity training set, stochastic generation binary combination and tlv triple, calculate respectively its correspondence contrast loss and
Tlv triple is lost, and skimming loss value is sent less than the combination of predetermined threshold value, a collection of training sample combination only retaining penalty values bigger
Enter model to be trained.
For low capacity training set, first calculate the similarity between all samples.Then minimum similar of similarity is chosen
Sample is as the positive sample of difficulty, and the highest inhomogeneity sample of similarity is as difficult negative sample, finally with the difficult positive sample filtered out
This is sent into model with difficult negative sample as training sample and is trained.
Along with the carrying out of network training, continuous regularized learning algorithm rate also screens difficult sample feeding training, until training loss is not
Reduce again, thus obtain final model.
Step S3-2, use step S3-1 train the neural network model obtained, the pretreatment that will obtain in step S2
After facial image as the input of neutral net, perform the forward calculation of neural network model, the neural network model obtained
Output be input facial image feature representation.
The facial image feature that step S4, calculation procedure S3 obtain and the similarity of facial image feature in data base, sentence
User identity in disconnected input facial image.
Case study on implementation:
In order to describe detailed description of the invention and the checking effectiveness of the invention of the present invention in detail, the present invention is proposed by we
Method be applied to a disclosed face database LFW face database.This data base includes 5749 people, altogether
13233 width images.
In our embodiment, we use the standard test protocols of LFW data set to prove effectiveness of the invention.
The standard test protocols of LFW data set is made up of 6000 pairs of facial images, wherein comprise the facial image of 3000 pairs of same person with
And the facial image of 3000 pairs of different people.
Specifically comprise the following steps that
Training process:
Step S3-1, collects a large amount of facial image as training data, design neural network model.Especially, Wo Mensuo
Use neural network model to comprise 4 convolutional layers and 4 pond layers, after the layer of each pond, output is divided into two groups, connect
The sequencing neural unit of big Value Operations.According to the order of step S3-1, softmax loss, contrast loss is used to damage with tlv triple
Lose and as optimization object function, model is trained.Along with the carrying out of network training, continuous regularized learning algorithm rate also screens difficult sample
This feeding is trained, until training loss no longer reduces, thus obtains final model.
Test process:
Step S1, first we carry out Face datection and critical point detection to all input pictures, obtain all input figures
The face location information of picture and key point positional information.
Step S2, the face location information and the key point information that obtain according to previous step carry out attitude school to facial image
The pretreatment operation such as just, illumination balance.Specifically, for LFW data set, we use rotation and scaling will input face
Image rectification is to front face.
Step S3, the pretreated facial image that step S2 obtains, as the input of neutral net, performs neutral net
The forward calculation of model, obtains the feature representation of 6000 pairs of facial images.
Step S4, calculates 6000 to the COS distance of facial image as its similarity.Adjust similarity threshold,
To correct percent of pass, misclassification rate and discrimination under each threshold value.
| TPR@FPR=1% | 92.73% |
| TPR@FPR=0.1% | 84.83% |
| TPR@FPR=0 | 74.63% |
| Accuracy rate | 98.18% |
| The calculating time | 78ms |
| Parameter model size | 26.3MB |
Table 1 is present invention recognition accuracy on LFW data base
Table 1 illustrates our the method correct percent of pass when misclassification rate is 0,0.1%, 1%, and under optimal threshold
Discrimination.The storage size of the degree of depth learning model Parameter File needs that table 1 also illustrates us is special with single picture
Levy the extraction time used.The human face recognition model that performance is suitable in the world is compared, our method calculate speed faster, storage
Expense is less.
Particular embodiments described above, has been carried out the purpose of the present invention, technical scheme and beneficial effect the most in detail
Describe in detail bright, be it should be understood that the specific embodiment that the foregoing is only the present invention, be not limited to the present invention, all
Within the spirit and principles in the present invention, any modification, equivalent substitution and improvement etc. done, should be included in the guarantor of the present invention
Within the scope of protecting.
Claims (10)
1. a face identification method based on sequencing neural network model, it is characterised in that including:
Step S1, the image to be identified of reading input, detect the face location in image to be identified and key point positional information；
Step S2, according to described face location information and key point information, image to be identified is carried out pretreatment operation；
Step S3, by the input of pretreated image to be identified in sequencing neural network model, obtain the spy of image to be identified
Levy expression；
Step S4, the feature representation calculating image to be identified and the similarity of known facial image feature in data base, to identify
Image to be identified.
Face identification method based on sequencing neural network model the most according to claim 1, it is characterised in that step S1
Also include:
For image to be identified, application image recognizer detects face location information, and according to the face location information of gained,
Application image recognizer obtains the key point positional information of face；Wherein, the key point of face is pre-defined.
Face identification method based on sequencing neural network model the most according to claim 1, it is characterised in that described step
Pretreatment operation described in rapid S2 includes attitude updating and light correction；Described attitude updating comprises determining that the key of standard face
Point position and illumination condition, then according to described face location information and key point information by the face key point of input picture
Put and be aligned to standard face key point position, to reach to correct the purpose of human face posture；Wherein, the key of standard face can be pre-defined
Point position and illumination condition, or be directly used in training set and be calculated average face as standard face, it is then determined that it is crucial
Dot position information and illumination condition standard face；
The correction of described light includes by image processing algorithm, by the light change of facial image to consistent with standard face.
Face identification method based on sequencing neural network model the most according to claim 1, it is characterised in that described step
Described sequencing neural network model in rapid S3 includes sequencing neutral net unit, for obtaining the sequencing table between multiple input
Reach.
Face identification method based on sequencing neural network model the most according to claim 4, it is characterised in that described fixed
Sequence is expressed and is included maximum, minima, meansigma methods and difference or long-pending business.
6. the face identification method based on sequencing neural network model as described in any one of claim 1-5, it is characterised in that
Described sequencing neural network model trains acquisition as follows:
Utilize the training sample in training set, use Classification Loss function to classify as optimization aim, training convolutional neural networks
Model, in order to classify to the face in training sample；
Using the convolutional neural networks disaggregated model that trains as the model of pre-training, contrast loss function and tlv triple is used to damage
Lose function to continue described convolutional neural networks disaggregated model is trained.
7. face identification method based on sequencing neural network model as claimed in claim 6, it is characterised in that described classification
Loss function is softmax loss function, is expressed as:
Wherein, N is class number；X is input；y∈RN×1It it is the categorization vector representing output；Represent the classification that training obtains
Device,Represent the output of convolutional neural networks disaggregated model output layer i-th node.
Face identification method based on sequencing neural network model the most as claimed in claim 6, it is characterised in that described in make
Continue described convolutional neural networks disaggregated model is trained middle contrast with contrast loss function and/or tlv triple loss function
Loss function is expressed as:
Wherein, θ controls the parameter of between class distance difference, described tlv triple loss function such as following table in class in being contrast loss function
Show:
L=max (d (f (xa), f (xp))-d(f(xa)+α, f (xn)), 0)
Wherein f (x) is the output of model time last layer of the input of model output layer of described pre-training, i.e. pre-training；D (.) is
Distance function, xaIt is referred to as central sample, xpFor with central sample xaBelong to of a sort positive sample, xnFor with central sample xaBelong to
Inhomogeneous negative sample；During training, central sample xaFor the sample of random choose in training set, a is that tlv triple is damaged
The parameter of between class distance difference in control class in mistake function.
Face identification method based on sequencing neural network model the most as claimed in claim 6, it is characterised in that described in make
Continue described convolutional neural networks disaggregated model is trained middle screening with contrast loss function and/or tlv triple loss function
Go out the combination of difficult sample as training data, concretely comprising the following steps of described difficult sample combined sorting:
For Large Copacity training set, stochastic generation binary combination and tlv triple, calculate the contrast loss and/or three of its correspondence respectively
Tuple is lost, and skimming loss value, less than the combination of predetermined threshold value, utilizes the training sample remained combination to be trained；
For low capacity training set, first calculate the similarity between all training samples, then choose minimum similar of similarity
Training sample is as the positive sample of difficulty, and the highest inhomogeneity sample of similarity is as difficult negative sample, finally with the difficulty filtered out
Positive sample is trained as training sample with difficult negative sample.
10. a face identification device based on sequencing neural network model, it is characterised in that including:
Input module, for reading in the image to be identified of input, detects the face location in image to be identified and key point position
Information；
Pretreatment module, for carrying out pretreatment behaviour according to described face location information and key point information to image to be identified
Make；
Feature acquisition module, for by pretreated image to be identified input to sequencing neural network model, obtaining waiting to know
The feature representation of other image；
Identification module, for calculating feature representation and the similarity of known facial image feature in data base of image to be identified,
To identify image to be identified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610403028.9A CN106096538B (en) | 2016-06-08 | 2016-06-08 | Face identification method and device based on sequencing neural network model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610403028.9A CN106096538B (en) | 2016-06-08 | 2016-06-08 | Face identification method and device based on sequencing neural network model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106096538A true CN106096538A (en) | 2016-11-09 |
| CN106096538B CN106096538B (en) | 2019-08-23 |
Family
ID=57229125
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610403028.9A Active CN106096538B (en) | 2016-06-08 | 2016-06-08 | Face identification method and device based on sequencing neural network model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106096538B (en) |
Cited By (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106709528A (en) * | 2017-01-10 | 2017-05-24 | 深圳大学 | Method and device of vehicle reidentification based on multiple objective function deep learning |
| CN107016708A (en) * | 2017-03-24 | 2017-08-04 | 杭州电子科技大学 | A kind of image Hash coding method based on deep learning |
| CN107480785A (en) * | 2017-07-04 | 2017-12-15 | 北京小米移动软件有限公司 | The training method and device of convolutional neural networks |
| CN107679513A (en) * | 2017-10-20 | 2018-02-09 | 北京达佳互联信息技术有限公司 | Image processing method, device and server |
| CN107678845A (en) * | 2017-09-30 | 2018-02-09 | 广东欧珀移动通信有限公司 | Application program management-control method, device, storage medium and electronic equipment |
| CN107729873A (en) * | 2017-11-03 | 2018-02-23 | 北京奇虎科技有限公司 | Special object detection method, device and computing device in video file |
| CN107797831A (en) * | 2017-11-14 | 2018-03-13 | 广东欧珀移动通信有限公司 | Background application method for cleaning, device, storage medium and electronic equipment |
| CN107818314A (en) * | 2017-11-22 | 2018-03-20 | 北京达佳互联信息技术有限公司 | Face image processing method, device and server |
| CN107944363A (en) * | 2017-11-15 | 2018-04-20 | 北京达佳互联信息技术有限公司 | Face image processing process, system and server |
| CN108108807A (en) * | 2017-12-29 | 2018-06-01 | 北京达佳互联信息技术有限公司 | Learning-oriented image processing method, system and server |
| CN108171209A (en) * | 2018-01-18 | 2018-06-15 | 中科视拓（北京）科技有限公司 | A kind of face age estimation method that metric learning is carried out based on convolutional neural networks |
| CN108182394A (en) * | 2017-12-22 | 2018-06-19 | 浙江大华技术股份有限公司 | Training method, face identification method and the device of convolutional neural networks |
| CN108205661A (en) * | 2017-12-27 | 2018-06-26 | 浩云科技股份有限公司 | A kind of ATM abnormal human face detection based on deep learning |
| CN108229269A (en) * | 2016-12-31 | 2018-06-29 | 深圳市商汤科技有限公司 | Method for detecting human face, device and electronic equipment |
| WO2018153201A1 (en) * | 2017-02-22 | 2018-08-30 | 中兴通讯股份有限公司 | Deep learning training method and apparatus |
| CN108830201A (en) * | 2018-06-01 | 2018-11-16 | 平安科技（深圳）有限公司 | Acquisition methods, device, computer equipment and the storage medium of sample triple |
| CN108921791A (en) * | 2018-07-03 | 2018-11-30 | 苏州中科启慧软件技术有限公司 | Lightweight image super-resolution improved method based on adaptive important inquiry learning |
| CN108921019A (en) * | 2018-05-27 | 2018-11-30 | 北京工业大学 | A kind of gait recognition method based on GEI and TripletLoss-DenseNet |
| CN109190446A (en) * | 2018-07-06 | 2019-01-11 | 西北工业大学 | Pedestrian's recognition methods again based on triple focused lost function |
| CN109359541A (en) * | 2018-09-17 | 2019-02-19 | 南京邮电大学 | A kind of sketch face identification method based on depth migration study |
| WO2019037346A1 (en) * | 2017-08-25 | 2019-02-28 | 广州视源电子科技股份有限公司 | Method and device for optimizing human face picture quality evaluation model |
| CN109657612A (en) * | 2018-12-19 | 2019-04-19 | 苏州纳智天地智能科技有限公司 | A kind of quality-ordered system and its application method based on facial image feature |
| CN109977639A (en) * | 2018-10-26 | 2019-07-05 | 招商银行股份有限公司 | Identity identifying method, device and computer readable storage medium |
| CN110569737A (en) * | 2019-08-15 | 2019-12-13 | 深圳华北工控软件技术有限公司 | Face recognition deep learning method and face recognition acceleration camera |
| CN110674941A (en) * | 2019-09-25 | 2020-01-10 | 南开大学 | Data encryption transmission method and system based on neural network |
| CN110674881A (en) * | 2019-09-27 | 2020-01-10 | 长城计算机软件与系统有限公司 | Trademark image retrieval model training method, system, storage medium and computer equipment |
| CN110674688A (en) * | 2019-08-19 | 2020-01-10 | 深圳力维智联技术有限公司 | Face recognition model acquisition method, system and medium for video monitoring scene |
| CN110766152A (en) * | 2018-07-27 | 2020-02-07 | 富士通株式会社 | Method and apparatus for training deep neural networks |
| CN110866962A (en) * | 2019-11-20 | 2020-03-06 | 成都威爱新经济技术研究院有限公司 | Virtual portrait and expression synchronization method based on convolutional neural network |
| CN110889348A (en) * | 2019-11-15 | 2020-03-17 | 亚信科技（中国）有限公司 | Method and device for improving success rate of face recognition under complex light |
| CN111402027A (en) * | 2018-12-14 | 2020-07-10 | 马上消费金融股份有限公司 | Identity recognition method, commodity loan auditing method, device and terminal equipment |
| CN111582199A (en) * | 2020-05-12 | 2020-08-25 | 佛山市玖章智能科技有限公司 | Face recognition model training method and face recognition method |
| CN111723613A (en) * | 2019-03-20 | 2020-09-29 | 广州慧睿思通信息科技有限公司 | Face image data processing method, device, equipment and storage medium |
| WO2020238101A1 (en) * | 2019-05-27 | 2020-12-03 | 商汤集团有限公司 | Object detection method, apparatus, device and computer-readable storage medium |
| CN112115860A (en) * | 2020-09-18 | 2020-12-22 | 深圳市威富视界有限公司 | Face key point positioning method and device, computer equipment and storage medium |
| CN112232117A (en) * | 2020-09-08 | 2021-01-15 | 深圳微步信息股份有限公司 | Face recognition method, face recognition device and storage medium |
| CN112508130A (en) * | 2020-12-25 | 2021-03-16 | 商汤集团有限公司 | Clustering method and device, electronic equipment and storage medium |
| CN112818887A (en) * | 2021-02-08 | 2021-05-18 | 中国科学院自动化研究所 | Human body skeleton sequence behavior identification method based on unsupervised learning |
| CN113282718A (en) * | 2021-07-26 | 2021-08-20 | 北京快鱼电子股份公司 | Language identification method and system based on self-adaptive center anchor |
| WO2021203882A1 (en) * | 2020-04-07 | 2021-10-14 | 深圳市商汤科技有限公司 | Attitude detection and video processing method and apparatus, and electronic device and storage medium |
| WO2021218899A1 (en) * | 2020-04-30 | 2021-11-04 | 京东方科技集团股份有限公司 | Method for training facial recognition model, and method and apparatus for facial recognition |
| TWI752455B (en) * | 2019-11-11 | 2022-01-11 | 大陸商深圳市商湯科技有限公司 | Image classification model training method, image processing method, data classification model training method, data processing method, computer device, and storage medium |
| WO2022126464A1 (en) * | 2020-12-17 | 2022-06-23 | 深圳市优必选科技股份有限公司 | Face recognition method, robot, and storage medium |
| CN115424335A (en) * | 2022-11-03 | 2022-12-02 | 智慧眼科技股份有限公司 | Living body recognition model training method, living body recognition method and related equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102629320A (en) * | 2012-03-27 | 2012-08-08 | 中国科学院自动化研究所 | Ordinal measurement statistical description face recognition method based on feature level |
| CN105608450A (en) * | 2016-03-01 | 2016-05-25 | 天津中科智能识别产业技术研究院有限公司 | Heterogeneous face identification method based on deep convolutional neural network |
-
2016
- 2016-06-08 CN CN201610403028.9A patent/CN106096538B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102629320A (en) * | 2012-03-27 | 2012-08-08 | 中国科学院自动化研究所 | Ordinal measurement statistical description face recognition method based on feature level |
| CN105608450A (en) * | 2016-03-01 | 2016-05-25 | 天津中科智能识别产业技术研究院有限公司 | Heterogeneous face identification method based on deep convolutional neural network |
Non-Patent Citations (2)
| Title |
|---|
| XIANG WU, ET AL.: "A Lightened CNN for Deep Face Representation", 《COMPUTER SCIENCE》 * |
| 赵亚龙: "基于卷积神经网络的三维人脸识别研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (67)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11182591B2 (en) | 2016-12-31 | 2021-11-23 | Shenzhen Sensetime Technology Co, Ltd | Methods and apparatuses for detecting face, and electronic devices |
| CN108229269A (en) * | 2016-12-31 | 2018-06-29 | 深圳市商汤科技有限公司 | Method for detecting human face, device and electronic equipment |
| CN106709528A (en) * | 2017-01-10 | 2017-05-24 | 深圳大学 | Method and device of vehicle reidentification based on multiple objective function deep learning |
| WO2018153201A1 (en) * | 2017-02-22 | 2018-08-30 | 中兴通讯股份有限公司 | Deep learning training method and apparatus |
| CN107016708A (en) * | 2017-03-24 | 2017-08-04 | 杭州电子科技大学 | A kind of image Hash coding method based on deep learning |
| CN107016708B (en) * | 2017-03-24 | 2020-06-05 | 杭州电子科技大学 | Image hash coding method based on deep learning |
| CN107480785A (en) * | 2017-07-04 | 2017-12-15 | 北京小米移动软件有限公司 | The training method and device of convolutional neural networks |
| CN107480785B (en) * | 2017-07-04 | 2021-01-29 | 北京小米移动软件有限公司 | Convolutional neural network training method and device |
| WO2019037346A1 (en) * | 2017-08-25 | 2019-02-28 | 广州视源电子科技股份有限公司 | Method and device for optimizing human face picture quality evaluation model |
| CN107678845B (en) * | 2017-09-30 | 2020-03-10 | Oppo广东移动通信有限公司 | Application program control method and device, storage medium and electronic equipment |
| CN107678845A (en) * | 2017-09-30 | 2018-02-09 | 广东欧珀移动通信有限公司 | Application program management-control method, device, storage medium and electronic equipment |
| CN107679513A (en) * | 2017-10-20 | 2018-02-09 | 北京达佳互联信息技术有限公司 | Image processing method, device and server |
| CN107729873A (en) * | 2017-11-03 | 2018-02-23 | 北京奇虎科技有限公司 | Special object detection method, device and computing device in video file |
| CN107797831A (en) * | 2017-11-14 | 2018-03-13 | 广东欧珀移动通信有限公司 | Background application method for cleaning, device, storage medium and electronic equipment |
| CN107797831B (en) * | 2017-11-14 | 2021-06-01 | Oppo广东移动通信有限公司 | Background application cleaning method and device, storage medium and electronic equipment |
| CN107944363A (en) * | 2017-11-15 | 2018-04-20 | 北京达佳互联信息技术有限公司 | Face image processing process, system and server |
| CN107818314A (en) * | 2017-11-22 | 2018-03-20 | 北京达佳互联信息技术有限公司 | Face image processing method, device and server |
| CN108182394A (en) * | 2017-12-22 | 2018-06-19 | 浙江大华技术股份有限公司 | Training method, face identification method and the device of convolutional neural networks |
| CN108205661A (en) * | 2017-12-27 | 2018-06-26 | 浩云科技股份有限公司 | A kind of ATM abnormal human face detection based on deep learning |
| CN108108807A (en) * | 2017-12-29 | 2018-06-01 | 北京达佳互联信息技术有限公司 | Learning-oriented image processing method, system and server |
| CN108108807B (en) * | 2017-12-29 | 2020-06-02 | 北京达佳互联信息技术有限公司 | Learning type image processing method, system and server |
| CN108171209A (en) * | 2018-01-18 | 2018-06-15 | 中科视拓（北京）科技有限公司 | A kind of face age estimation method that metric learning is carried out based on convolutional neural networks |
| CN108171209B (en) * | 2018-01-18 | 2021-07-16 | 中科视拓（北京）科技有限公司 | Face age estimation method for metric learning based on convolutional neural network |
| CN108921019B (en) * | 2018-05-27 | 2022-03-08 | 北京工业大学 | Gait recognition method based on GEI and TripletLoss-DenseNet |
| CN108921019A (en) * | 2018-05-27 | 2018-11-30 | 北京工业大学 | A kind of gait recognition method based on GEI and TripletLoss-DenseNet |
| CN108830201A (en) * | 2018-06-01 | 2018-11-16 | 平安科技（深圳）有限公司 | Acquisition methods, device, computer equipment and the storage medium of sample triple |
| CN108830201B (en) * | 2018-06-01 | 2020-06-23 | 平安科技（深圳）有限公司 | Method and device for acquiring sample triple, computer equipment and storage medium |
| CN108921791A (en) * | 2018-07-03 | 2018-11-30 | 苏州中科启慧软件技术有限公司 | Lightweight image super-resolution improved method based on adaptive important inquiry learning |
| CN109190446A (en) * | 2018-07-06 | 2019-01-11 | 西北工业大学 | Pedestrian's recognition methods again based on triple focused lost function |
| CN110766152A (en) * | 2018-07-27 | 2020-02-07 | 富士通株式会社 | Method and apparatus for training deep neural networks |
| CN110766152B (en) * | 2018-07-27 | 2023-08-04 | 富士通株式会社 | Method and apparatus for training deep neural networks |
| CN109359541A (en) * | 2018-09-17 | 2019-02-19 | 南京邮电大学 | A kind of sketch face identification method based on depth migration study |
| CN109977639A (en) * | 2018-10-26 | 2019-07-05 | 招商银行股份有限公司 | Identity identifying method, device and computer readable storage medium |
| CN109977639B (en) * | 2018-10-26 | 2021-05-04 | 招商银行股份有限公司 | Identity authentication method and device and computer readable storage medium |
| CN111402027A (en) * | 2018-12-14 | 2020-07-10 | 马上消费金融股份有限公司 | Identity recognition method, commodity loan auditing method, device and terminal equipment |
| CN111402027B (en) * | 2018-12-14 | 2021-05-07 | 马上消费金融股份有限公司 | Identity recognition method, commodity loan auditing method, device and terminal equipment |
| CN109657612B (en) * | 2018-12-19 | 2023-12-12 | 苏州纳智天地智能科技有限公司 | Quality sorting system based on facial image features and application method thereof |
| CN109657612A (en) * | 2018-12-19 | 2019-04-19 | 苏州纳智天地智能科技有限公司 | A kind of quality-ordered system and its application method based on facial image feature |
| CN111723613A (en) * | 2019-03-20 | 2020-09-29 | 广州慧睿思通信息科技有限公司 | Face image data processing method, device, equipment and storage medium |
| JP7101267B2 (en) | 2019-05-27 | 2022-07-14 | センスタイム グループ リミテッド | Target detection method, device, equipment and computer readable storage medium |
| KR20210006429A (en) * | 2019-05-27 | 2021-01-18 | 센스타임 그룹 리미티드 | Object detection method, apparatus, equipment and computer-readable storage medium |
| JP2021530015A (en) * | 2019-05-27 | 2021-11-04 | センスタイム グループ リミテッド | Target detection method, device, device and computer readable storage medium |
| KR102455745B1 (en) * | 2019-05-27 | 2022-10-17 | 센스타임 그룹 리미티드 | Object detection method, apparatus, equipment and computer readable storage medium |
| WO2020238101A1 (en) * | 2019-05-27 | 2020-12-03 | 商汤集团有限公司 | Object detection method, apparatus, device and computer-readable storage medium |
| CN110569737A (en) * | 2019-08-15 | 2019-12-13 | 深圳华北工控软件技术有限公司 | Face recognition deep learning method and face recognition acceleration camera |
| CN110674688A (en) * | 2019-08-19 | 2020-01-10 | 深圳力维智联技术有限公司 | Face recognition model acquisition method, system and medium for video monitoring scene |
| CN110674688B (en) * | 2019-08-19 | 2023-10-31 | 深圳力维智联技术有限公司 | Face recognition model acquisition method, system and medium for video monitoring scene |
| CN110674941A (en) * | 2019-09-25 | 2020-01-10 | 南开大学 | Data encryption transmission method and system based on neural network |
| CN110674881A (en) * | 2019-09-27 | 2020-01-10 | 长城计算机软件与系统有限公司 | Trademark image retrieval model training method, system, storage medium and computer equipment |
| CN110674881B (en) * | 2019-09-27 | 2022-02-11 | 长城计算机软件与系统有限公司 | Trademark image retrieval model training method, system, storage medium and computer equipment |
| TWI752455B (en) * | 2019-11-11 | 2022-01-11 | 大陸商深圳市商湯科技有限公司 | Image classification model training method, image processing method, data classification model training method, data processing method, computer device, and storage medium |
| CN110889348A (en) * | 2019-11-15 | 2020-03-17 | 亚信科技（中国）有限公司 | Method and device for improving success rate of face recognition under complex light |
| CN110866962B (en) * | 2019-11-20 | 2023-06-16 | 成都威爱新经济技术研究院有限公司 | Virtual portrait and expression synchronization method based on convolutional neural network |
| CN110866962A (en) * | 2019-11-20 | 2020-03-06 | 成都威爱新经济技术研究院有限公司 | Virtual portrait and expression synchronization method based on convolutional neural network |
| WO2021203882A1 (en) * | 2020-04-07 | 2021-10-14 | 深圳市商汤科技有限公司 | Attitude detection and video processing method and apparatus, and electronic device and storage medium |
| WO2021218899A1 (en) * | 2020-04-30 | 2021-11-04 | 京东方科技集团股份有限公司 | Method for training facial recognition model, and method and apparatus for facial recognition |
| CN111582199A (en) * | 2020-05-12 | 2020-08-25 | 佛山市玖章智能科技有限公司 | Face recognition model training method and face recognition method |
| CN111582199B (en) * | 2020-05-12 | 2023-05-23 | 佛山市玖章智能科技有限公司 | Face recognition model training method and face recognition method |
| CN112232117A (en) * | 2020-09-08 | 2021-01-15 | 深圳微步信息股份有限公司 | Face recognition method, face recognition device and storage medium |
| CN112115860A (en) * | 2020-09-18 | 2020-12-22 | 深圳市威富视界有限公司 | Face key point positioning method and device, computer equipment and storage medium |
| WO2022126464A1 (en) * | 2020-12-17 | 2022-06-23 | 深圳市优必选科技股份有限公司 | Face recognition method, robot, and storage medium |
| CN112508130A (en) * | 2020-12-25 | 2021-03-16 | 商汤集团有限公司 | Clustering method and device, electronic equipment and storage medium |
| CN112818887B (en) * | 2021-02-08 | 2023-04-07 | 中国科学院自动化研究所 | Human skeleton sequence behavior identification method based on unsupervised learning |
| CN112818887A (en) * | 2021-02-08 | 2021-05-18 | 中国科学院自动化研究所 | Human body skeleton sequence behavior identification method based on unsupervised learning |
| CN113282718A (en) * | 2021-07-26 | 2021-08-20 | 北京快鱼电子股份公司 | Language identification method and system based on self-adaptive center anchor |
| CN115424335A (en) * | 2022-11-03 | 2022-12-02 | 智慧眼科技股份有限公司 | Living body recognition model training method, living body recognition method and related equipment |
| CN115424335B (en) * | 2022-11-03 | 2023-08-04 | 智慧眼科技股份有限公司 | Living body recognition model training method, living body recognition method and related equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106096538B (en) | 2019-08-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106096538B (en) | Face identification method and device based on sequencing neural network model | |
| CN106815566B (en) | Face retrieval method based on multitask convolutional neural network | |
| CN107273845B (en) | Facial expression recognition method based on confidence region and multi-feature weighted fusion | |
| CN106096535B (en) | Face verification method based on bilinear joint CNN | |
| CN104143079B (en) | The method and system of face character identification | |
| CN104050471B (en) | Natural scene character detection method and system | |
| CN109492529A (en) | A kind of Multi resolution feature extraction and the facial expression recognizing method of global characteristics fusion | |
| CN108268859A (en) | A kind of facial expression recognizing method based on deep learning | |
| CN106548159A (en) | Reticulate pattern facial image recognition method and device based on full convolutional neural networks | |
| CN109101938B (en) | Multi-label age estimation method based on convolutional neural network | |
| CN105550658A (en) | Face comparison method based on high-dimensional LBP (Local Binary Patterns) and convolutional neural network feature fusion | |
| CN102270308B (en) | Facial feature location method based on five sense organs related AAM (Active Appearance Model) | |
| CN105809123A (en) | Face detecting method and device | |
| CN109815826A (en) | The generation method and device of face character model | |
| CN104361328A (en) | Facial image normalization method based on self-adaptive multi-column depth model | |
| CN105005765A (en) | Facial expression identification method based on Gabor wavelet and gray-level co-occurrence matrix | |
| CN105447441A (en) | Face authentication method and device | |
| CN111401145B (en) | Visible light iris recognition method based on deep learning and DS evidence theory | |
| CN104091169A (en) | Behavior identification method based on multi feature fusion | |
| CN110781829A (en) | Light-weight deep learning intelligent business hall face recognition method | |
| CN104504383B (en) | A kind of method for detecting human face based on the colour of skin and Adaboost algorithm | |
| CN104834941A (en) | Offline handwriting recognition method of sparse autoencoder based on computer input | |
| CN109344856B (en) | Offline signature identification method based on multilayer discriminant feature learning | |
| CN104915684B (en) | A kind of image-recognizing method and device based on the more plane SVMs of robust | |
| CN110781897A (en) | Semantic edge detection method based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |