CN109190518A - A kind of face verification method based on general set metric learning - Google Patents
A kind of face verification method based on general set metric learning Download PDFInfo
- Publication number
- CN109190518A CN109190518A CN201810925973.4A CN201810925973A CN109190518A CN 109190518 A CN109190518 A CN 109190518A CN 201810925973 A CN201810925973 A CN 201810925973A CN 109190518 A CN109190518 A CN 109190518A
- Authority
- CN
- China
- Prior art keywords
- image
- face
- indicate
- measures
- dispersion
- 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
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
Abstract
The invention discloses a kind of face verification methods based on general set metric learning, belong to face verification field, to the facial image in known human face data collection α and unknown face data set β, the feature for extracting face respectively measures the measures of dispersion in data set α and data set β using aggregation degree；Distance metric function and distance metric are obtained using the measures of dispersion；Decision function is constructed using the distance metric function and distance metric, and using the minimum value for intersecting empirical risk function in gradient descent algorithm solution building decision function, the corresponding known facial image of the minimum value of the empirical risk function is the verification result of the unknown facial image；The present invention can be improved the accuracy of face verification.
Description
Technical field
The present invention relates to face verification fields, and in particular to a kind of face verification side based on general set metric learning
Method.
Background technique
With the development of science and technology and the improvement of people's living standards, Face recognition technology have also obtained extensive research
With exploitation, this recognition of face also becomes one of research theme most popular in Pattern recognition and image processing in nearly 30 years.It is so-called
Recognition of face is exactly to analyze face video or image using computer, and scheme to take out effective face characteristic identification information etc.,
And finally judge the identity of face object.In general, recognition of face problem is macroscopically divided into two classes: recognition of face and face
Verifying.Wherein, the one-to-many comparison of the thing that recognition of face is done, that is, judge the people that system is currently seen, for the everybody met in advance
Which of, such as access control system, attendance is registered and suspect tracks etc., and what face verification was done is one-to-one comparison,
Namely judge whether the people in two pictures is same people, and the most common application scenarios are exactly face unlock, by using terminal
Equipment compares the photo of the photo of user's registration in advance and collection in worksite, judges whether it is same people, identity can be completed
Verifying.
It is more and more applied to face recognition technology in our daily life recent years, people are to people
The requirement of face accuracy of identification is also higher and higher, and at the same time, face verification technology also has become the weight in field of face identification
Research direction is wanted, and face verification problem encountered also results in the interest and research of numerous scientific research personnel.In human face data
It concentrates, often encounters the variation of low, in different scenes the intensity of illumination of target facial image resolution ratio, the change of scale and visual angle
Change, the variation of human face expression movement, the variation of monitoring camera head apparatus etc. all can cause the same face in different scenes
Appearance the change of divergence it is huge, this is but also face verification technology encounters challenge.
Summary of the invention
It is an object of the invention to: a kind of face verification method based on general set metric learning is provided, is solved
At present because of the variation of intensity of illumination, scale and the variation at visual angle, the variation of human face expression posture and different prisons in face verification
The device configuration of control camera changes the more low technical problem of accuracy of face verification caused by a series of problems.
The technical solution adopted by the invention is as follows:
A kind of face verification method based on general set metric learning, comprising the following steps:
Step 1: to the facial image in known human face data collection α and unknown face data set β, extracting the spy of face respectively
Sign measures the measures of dispersion in data set α and data set β using aggregation degree；
Step 2: obtaining distance metric function and distance metric using the measures of dispersion；
Step 3: constructing decision function using the distance metric function and distance metric, and using under intersection gradient
The minimum value that algorithm solves empirical risk function in building decision function drops, and the minimum value of the empirical risk function is corresponding
Know that facial image is the verification result of the unknown facial image.
It further, further include being pre-processed to facial image in the step 1, the pretreatment includes to face
Face in image carries out mark frame processing, and is aligned facial image by rotation process.
Further, the expression formula of measures of dispersion is as follows in the step 1:
Fi α=[f1 α-c1 α；f2 α-c2 α；...；fNr α-cNr α] (1),
Fi β=[f1 β-c1 β；f2 β-c2 β；...；fNr β-cNr β] (2),
Wherein, fi αIndicate the feature vector of i-th of image in known human face data collection α, fi βUnknown face data set β
In i-th of image feature vector,Indicate the measures of dispersion of i-th of image in known human face data collection α,Indicate the measures of dispersion of i-th of image in unknown face data set β, NfIndicate the number of characteristics of image, NrIt indicates
The number of image, { c in reference set1 α, c2 α..., cNr αIndicate facial image feature in known human face data collection α, { c1 β,
c2 β..., cNr βIndicate facial image feature in unknown face data set β.
Further, in the step 2 distance metric function expression formula are as follows:
Wherein,Indicate the premultiplication quantum of interaction difference projection amount,Indicate that interaction difference is thrown
The right side of shadow amount multiplies quantum, and F indicates to seek the arithmetic square root of all elements quadratic sum.
Further, in the step 2 distance metric expression formula are as follows:
Wherein, biIndicate the distance metric of i-th of image, S indicates similar pair of set, and D indicates dissimilar right
Set,Indicate the measures of dispersion of i-th of image in known human face data collection α,It indicates not
Know the measures of dispersion of i-th of image in human face data collection β, NfIndicate the number of characteristics of image, NrIndicate the number of the image in reference set
Mesh.
Further, in the step 3, the expression formula of decision function are as follows:
f(Fi α, Fi β；L, R)=T-dL, R(Fi α, Fi β) (5),
Wherein, T indicates global decisions threshold value；
The expression formula of empirical risk function are as follows:
Wherein, wiIt is the weight of i-th of image difference amount.
In conclusion by adopting the above-described technical solution, the beneficial effects of the present invention are:
1. in order to allow face verification method to become more efficiently, and further improving its accuracy rate, present invention innovation
A kind of effective logic discrimination criterion has been used to property, has gone design one using initial data and characteristic information in the training process
The decision rule of a local auto-adaptive, by using the method for logic judgment so that output only 1 and -1 the two as a result, from
And calculation amount is considerably reduced to a certain extent, arithmetic speed is accelerated, also let us becomes when analyzing result
It is more clear understandable.
2. the present invention is traditional on the basis of constructing distance metric expression formula using vector metric, innovative proposition
With set measurement come the method for replacing vector metric, not only introduces measures of dispersion and describe Fi αAnd Fi β, also introduce interactive difference
Projection amount L multiplies quantum as the right side as premultiplication quantum, R, the two projector quantums L and R together constitute set measurement, allows L pairs
Each difference has effect, and R works to different difference, difference can be allowed to become readily apparent from, namely allows between identical face
Distance reduce as far as possible and the distance between different faces allowed to increase as far as possible, just because of it is such a based on collection
The module for closing metric learning, just can intuitively judge as a result, also while to a certain extent improving accuracy rate very much.
The parameter of L and limitation are also fewer so simultaneously, also facilitate addition constraint appropriate as the case may be, to face
The raising of verification technique produces certain positive influence, is also more conducive to being applied in our daily life.
3. when extracting feature, due to the variation at the change of intensity of illumination, scale and visual angle, human face posture expression
When variation (for example laugh heartily, cry bitter tears, grimace etc.), camera shooting appointed condition it is unconventionally graceful a series of problems, such as all
It will cause each species diversity of same face at different conditions, it is correct when face verification so as to cause face authentication failed is carried out
The problems such as rate is lower, in response to this, this method describe sub- F using set measurement by studyi αWith carry out Fi βCarry out measures of dispersion
Description, can not only describe than the vector characteristics used before more efficient, also can greatly weaken due to environmental factor, field in this way
Scape factor or camera conditions etc. change brought some adverse effects, so that also let us can be carried out targetedly
The feature of target facial image describes and the operations such as extraction.
Detailed description of the invention
Examples of the present invention will be described by way of reference to the accompanying drawings, in which:
Fig. 1 is arrangement flow chart of the invention；
Fig. 2 is the relationship building process expression figure of measures of dispersion and projection amount in the present invention；
Fig. 3 is the process comparison diagram for being transformed into the distance metric based on set measurement in the present invention from vector distance measurement；
Fig. 4 is the explanatory diagram in the present invention in relation to gathering metric learning.
Specific embodiment
All features disclosed in this specification or disclosed all methods or in the process the step of, in addition to mutually exclusive
Feature and/or step other than, can combine in any way.
It elaborates below with reference to Fig. 1-4 couples of present invention.
A kind of face verification method based on general set metric learning, comprising the following steps:
Step 1: to the facial image in known human face data collection α and unknown face data set β, extracting the spy of face respectively
Sign measures the measures of dispersion in data set α and data set β using aggregation degree；
Step 2: obtaining distance metric function and distance metric using the measures of dispersion；
Step 3: constructing decision function using the distance metric function and distance metric, and using under intersection gradient
The minimum value that algorithm solves empirical risk function in building decision function drops, and the minimum value of the empirical risk function is corresponding
Know that facial image is the verification result of the unknown facial image.
It further, further include being pre-processed to facial image in the step 1, the pretreatment includes to face
Face in image carries out mark frame processing, and is aligned facial image by rotation process.
Further, the expression formula of measures of dispersion is as follows in the step 1:
Fi α=[f1 α-c1 α；f2 α-c2 α；...；fNr α-cNr α] (8),
Fi β=[f1 β-c1 β；f2 β-c2 β；...；fNr β-cNr β] (9),
Wherein, fi αIndicate the feature vector of i-th of image in known human face data collection α, fi βUnknown face data set β
In i-th of image feature vector,Indicate the measures of dispersion of i-th of image in known human face data collection α,Indicate the measures of dispersion of i-th of image in unknown face data set β, NfIndicate the number of characteristics of image, NrIt indicates
The number of image, { c in reference set1 α, c2 α..., cNr αIndicate facial image feature in known human face data collection α, { c1 β,
c2 β..., cNr βIndicate facial image feature in unknown face data set β.
Further, in the step 2 distance metric function expression formula are as follows:
Wherein,Indicate the premultiplication quantum of interaction difference projection amount,Indicate that interaction difference is thrown
The right side of shadow amount multiplies quantum, and F indicates to seek the arithmetic square root of all elements quadratic sum.
Further, in the step 2 distance metric expression formula are as follows:
Wherein, biIndicate the distance metric of i-th of image, S indicates similar pair of set, and D indicates dissimilar right
Set,Indicate the measures of dispersion of i-th of image in known human face data collection α,It indicates not
Know the measures of dispersion of i-th of image in human face data collection β, NfIndicate the number of characteristics of image, NrIndicate the number of the image in reference set
Mesh.
Further, in the step 3, the expression formula of decision function are as follows:
f(Fi α, Fi β；L, R)=T-dL, R(Fi α, Fi β) (12),
Wherein, T indicates global decisions threshold value；
The expression formula of empirical risk function are as follows:
Wherein, wiIt is the weight of i-th of image difference amount.
Specific embodiment
As shown in Figure 1, this method includes: to carry out feature extraction firstly the need of to acquired image, in feature space
By study using set measurement Fi αAnd Fi βThe description of measures of dispersion is carried out, by transformation by vector fi αAnd fi βTransformation is for difference
Different amount Fi αAnd FiShown in β, measures of dispersion and its distance are defined as follows:
Fi α=[f1 α-c1 α；f2 α-c2 α；...；fNr α-cNr α] (15)
Fi β=[f1 β-c1 β；f2 β-c2 β；...；fNr β-cNr β] (16)
Fi α-Fi β
=[(f1 α-f1 β)-(c1 α-c1 β)；(f2 α-f2 β)-(c2 α-c2 β)；...；(fNr α-fNr β)-(cNr α-cNr β)]
(17)
In the present invention, two groups of reference image data collection have been selected, wherein { c1 α, c2 α..., cNr αIndicate to come from data
Collect the facial image feature description of α；And { c1 β, c2 β..., cNr βIndicate the facial image feature description from data set β；(fi α, fi β) then it is characterized the vector form of description；(Fi α, Fi β) then it is characterized the matrix form of description；And Fi α-Fiβ then indicates difference
The distance of amount, to also facilitate our subsequent sequence of operations to a certain extent.
Interaction difference projection amount L is introduced as premultiplication quantum, R multiplies quantum as the right side, and by the two projector quantums L and R
Set measurement is together constituted, to construct the distance metric representation d based on matrixL, R(Fi α, Fi β), specifically away from
It is as follows from measurement expression formula:
Construct decision function f (Fi α, Fi β；L, R), if when f > 0, illustrating that two related faces are similar；Conversely, working as f
≤ 0, explanation is dissimilar.Specific representation is as follows:
f(Fi α, Fi β；L, R)=T-dL, R(Fi α, Fi β) (19)
Wherein, T is mainly used to the distance between comparison two related faces as a global decisions threshold value, thus certainly
Determine whether they are similar.
Finally, constantly iteratively solving objective function J (L, R) by using the method for intersecting gradient decline, obtaining one most
Small empiric risk value, the expression formula of this empiric risk are as follows:
Wherein, f is a decision function,It is a loss function, and is monotone decreasing；wiIt is i-th pair difference
Measure the weight of description.
Gradient is solved to objective function J (L, R), as follows:
Decline by constantly carrying out intersection gradient in this way, is iterated solution, until reaching set threshold value, thus
Optimal solution is acquired, the smallest empiric risk value is obtained, the smallest known image of empiric risk value is the verifying knot of the unknown images
Fruit.
Fig. 2 is that building process of the present invention in relation to measures of dispersion indicates figure, it is contemplated that the validity of premultiplication is poor at each
It is different to increase weight above and the right validity multiplied is worked to different difference.Not with the vector metric method that used in the past
With this method uses set measurement, introduces interaction difference projection amountIt is sub as premultiplication,As right multiplier, the two projector quantums L and R together constitute set measurement.Wherein, it means that L is to every
A difference has effect, and R works to different difference.And as shown in connection with fig. 2, in measures of dispersion, based on set measurement phase
The principle multiplied, the row of L and the column of measures of dispersion be combined with each other, and similarly, the column of R and the row of measures of dispersion be combined with each other, this also illustrates
The relationship building process of measures of dispersion and two projector quantums L and R.
Fig. 3 is from vector distance measurement to the conversion process comparison diagram of the distance metric based on set measurement in the present invention.
For the face picture information in two different data sets, comparing the method that traditional method is proposed with us can be sent out
It is existing, in feature description and distance metric, this method relatively before method have very big improvement: feature description is come
It says, by it from vector form (fi α, fi β) it has been transformed into matrix form (Fi α, Fiβ), and by original Projection Character L it is converted to
Internal diversity projects L and R, so that by the original distance metric expression formula d based on vector formL(fi α, fi β) become
Distance metric expression formula d based on set measurementL, R(Fi α, Fi β).As shown in Figure 3, wherein NfRefer to the dimension of feature vector
Number, NrRefer to the number of reference picture set.
Fig. 4 is the explanatory diagram of the related set metric learning proposed in the present invention, has been largely divided into two large divisions, one is
Continuous item, it is mainly exactly that the combination of the measures of dispersion of the same face is allowed to draw closer together, and the other is outlier, it is mainly
Be allow different faces measures of dispersion combination separation more open.This also just constitutes the basic mistake in relation to gathering metric learning
Journey.
Above-described specific implementation is only a kind of best implementation of the invention, and what is be not intended to restrict the invention is special
Sharp range, it is all using equivalent structure or equivalent flow shift made by spirit of that invention and principle and accompanying drawing content, it should all wrap
It includes in scope of patent protection of the invention.
Claims (6)
1. a kind of face verification method based on general set metric learning, it is characterised in that: the following steps are included:
Step 1: to the facial image in known human face data collection α and unknown face data set β, the feature of face is extracted respectively,
The measures of dispersion in data set α and data set β is measured using aggregation degree；
Step 2: obtaining distance metric function and distance metric using the measures of dispersion；
Step 3: constructing decision function using the distance metric function and distance metric, and calculated using gradient decline is intersected
Method solves the minimum value of empirical risk function in building decision function, the corresponding known people of the minimum value of the empirical risk function
Face image is the verification result of the unknown facial image.
2. a kind of face verification method based on general set metric learning according to claim 1, it is characterised in that:
It further include being pre-processed to facial image in the step 1, the pretreatment includes marking to the face in facial image
Frame processing, and facial image is aligned by rotation process.
3. a kind of face verification method based on general set metric learning according to claim 1, it is characterised in that:
The expression formula of measures of dispersion is as follows in the step 1:
Fi α=[f1 α-c1 α；f2 α-c2 α；...；fNr α-cNr α] (1),
Fi β=[f1 β-c1 β；f2 β-c2 β；...；fNr β-cNr β] (2),
Wherein, fi αIndicate the feature vector of i-th of image in known human face data collection α, fi βI-th in unknown face data set β
The feature vector of a image,Indicate the measures of dispersion of i-th of image in known human face data collection α,Indicate the measures of dispersion of i-th of image in unknown face data set β, NfIndicate the number of characteristics of image, NrIt indicates
In the number of known human face data collection image, { c1α, c2α ..., cNr αIndicate that the facial image in known human face data collection α is special
Sign, { c1 β, c2 β..., cNr βIndicate facial image feature in unknown face data set β.
4. a kind of face verification method based on general set metric learning according to claim 1, it is characterised in that:
The expression formula of distance metric function in the step 2 are as follows:
Wherein,Indicate the premultiplication quantum of interaction difference projection amount,Indicate interaction difference projection amount
The right side multiply quantum, F indicates to seek the arithmetic square root of all elements quadratic sum.
5. a kind of face verification method based on general set metric learning according to claim 1, it is characterised in that:
The expression formula of distance metric in the step 2 are as follows:
Wherein, biIndicate the distance metric of i-th of image, S indicates similar pair of set, and D indicates dissimilar pair of collection
It closes,Indicate the measures of dispersion of i-th of image in known human face data collection α,Indicate unknown human
The measures of dispersion of i-th of image, N in face data set βfIndicate the number of characteristics of image, NrIndicate the number of the image in reference set.
6. a kind of face verification method based on general set metric learning according to claim 1, it is characterised in that:
In the step 3, the expression formula of decision function are as follows:
f(Fi α, Fi β；L, R)=T-dL, R(Fi α, Fi β) (5),
Wherein, T indicates global decisions threshold value；
The expression formula of empirical risk function are as follows:
Wherein, wiIt is the weight of i-th of image difference amount.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810925973.4A CN109190518B (en) | 2018-08-14 | 2018-08-14 | Face verification method based on universal set metric learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810925973.4A CN109190518B (en) | 2018-08-14 | 2018-08-14 | Face verification method based on universal set metric learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109190518A true CN109190518A (en) | 2019-01-11 |
| CN109190518B CN109190518B (en) | 2022-03-18 |
Family
ID=64921794
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810925973.4A Active CN109190518B (en) | 2018-08-14 | 2018-08-14 | Face verification method based on universal set metric learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109190518B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102592148A (en) * | 2011-12-29 | 2012-07-18 | 华南师范大学 | Face identification method based on non-negative matrix factorization and a plurality of distance functions |
| US8233702B2 (en) * | 2006-08-18 | 2012-07-31 | Google Inc. | Computer implemented technique for analyzing images |
| CN104765768A (en) * | 2015-03-09 | 2015-07-08 | 深圳云天励飞技术有限公司 | Mass face database rapid and accurate retrieval method |
| US20150286638A1 (en) * | 2012-11-09 | 2015-10-08 | Orbeus, Inc. | System, method and apparatus for scene recognition |
| CN105678260A (en) * | 2016-01-07 | 2016-06-15 | 浙江工贸职业技术学院 | Sparse maintenance distance measurement-based human face identification method |
| CN106599833A (en) * | 2016-12-12 | 2017-04-26 | 武汉科技大学 | Field adaptation and manifold distance measurement-based human face identification method |
| CN107657223A (en) * | 2017-09-18 | 2018-02-02 | 华南理工大学 | It is a kind of based on the face authentication method for quickly handling more learning distance metrics |
-
2018
- 2018-08-14 CN CN201810925973.4A patent/CN109190518B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8233702B2 (en) * | 2006-08-18 | 2012-07-31 | Google Inc. | Computer implemented technique for analyzing images |
| CN102592148A (en) * | 2011-12-29 | 2012-07-18 | 华南师范大学 | Face identification method based on non-negative matrix factorization and a plurality of distance functions |
| US20150286638A1 (en) * | 2012-11-09 | 2015-10-08 | Orbeus, Inc. | System, method and apparatus for scene recognition |
| CN104765768A (en) * | 2015-03-09 | 2015-07-08 | 深圳云天励飞技术有限公司 | Mass face database rapid and accurate retrieval method |
| CN105678260A (en) * | 2016-01-07 | 2016-06-15 | 浙江工贸职业技术学院 | Sparse maintenance distance measurement-based human face identification method |
| CN106599833A (en) * | 2016-12-12 | 2017-04-26 | 武汉科技大学 | Field adaptation and manifold distance measurement-based human face identification method |
| CN107657223A (en) * | 2017-09-18 | 2018-02-02 | 华南理工大学 | It is a kind of based on the face authentication method for quickly handling more learning distance metrics |
Non-Patent Citations (3)
| Title |
|---|
| F. WANG 等: "Regularizing face verification nets for pain intensity regression", 《2017 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING》 * |
| PANAGIOTIS MOUTAFIS 等: "An Overview and Empirical Comparison of Distance Metric Learning Methods", 《IEEE TRANSACTIONS ON CYBERNETICS》 * |
| 吴嘉琪 等: "基于集成人脸对距离学习的跨年龄人脸验证", 《模式识别与人工智能》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109190518B (en) | 2022-03-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106845357B (en) | A kind of video human face detection and recognition methods based on multichannel network | |
| CN108537743B (en) | Face image enhancement method based on generation countermeasure network | |
| CN106503687B (en) | Merge the monitor video system for identifying figures and its method of face multi-angle feature | |
| CN108764308B (en) | Pedestrian re-identification method based on convolution cycle network | |
| CN104166841B (en) | The quick detection recognition methods of pedestrian or vehicle is specified in a kind of video surveillance network | |
| CN104866829B (en) | A kind of across age face verification method based on feature learning | |
| CN110889672B (en) | Student card punching and class taking state detection system based on deep learning | |
| CN109033938A (en) | A kind of face identification method based on ga s safety degree Fusion Features | |
| CN108875600A (en) | A kind of information of vehicles detection and tracking method, apparatus and computer storage medium based on YOLO | |
| US9639747B2 (en) | Online learning method for people detection and counting for retail stores | |
| CN108229330A (en) | Face fusion recognition methods and device, electronic equipment and storage medium | |
| CN107463920A (en) | A kind of face identification method for eliminating partial occlusion thing and influenceing | |
| CN109190475B (en) | Face recognition network and pedestrian re-recognition network collaborative training method | |
| CN108921051A (en) | Pedestrian's Attribute Recognition network and technology based on Recognition with Recurrent Neural Network attention model | |
| CN106447625A (en) | Facial image series-based attribute identification method and device | |
| CN107895160A (en) | Human face detection and tracing device and method | |
| CN109063649A (en) | Pedestrian's recognition methods again of residual error network is aligned based on twin pedestrian | |
| CN109635634A (en) | A kind of pedestrian based on stochastic linear interpolation identifies data enhancement methods again | |
| CN103390151B (en) | Method for detecting human face and device | |
| CN104298974A (en) | Human body behavior recognition method based on depth video sequence | |
| CN106096517A (en) | A kind of face identification method based on low-rank matrix Yu eigenface | |
| CN109190472A (en) | Combine pedestrian's attribute recognition approach of guidance with attribute based on image | |
| CN113205002B (en) | Low-definition face recognition method, device, equipment and medium for unlimited video monitoring | |
| CN105893947A (en) | Bi-visual-angle face identification method based on multi-local correlation characteristic learning | |
| CN111368768A (en) | Human body key point-based employee gesture guidance detection method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |