CN108765014A - A kind of intelligent advertisement put-on method based on access control system - Google Patents
A kind of intelligent advertisement put-on method based on access control system Download PDFInfo
- Publication number
- CN108765014A CN108765014A CN201810540510.6A CN201810540510A CN108765014A CN 108765014 A CN108765014 A CN 108765014A CN 201810540510 A CN201810540510 A CN 201810540510A CN 108765014 A CN108765014 A CN 108765014A
- Authority
- CN
- China
- Prior art keywords
- age
- visitor
- parameter
- face
- identification
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0252—Targeted advertisements based on events or environment, e.g. weather or festivals
-
- 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
-
- 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
- G06V40/171—Local features and components; Facial parts ; Occluding parts, e.g. glasses; Geometrical relationships
-
- 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
- G06V40/178—Human faces, e.g. facial parts, sketches or expressions estimating age from face image; using age information for improving recognition
Landscapes
- Engineering & Computer Science (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Accounting & Taxation (AREA)
- Strategic Management (AREA)
- Human Computer Interaction (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Multimedia (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Game Theory and Decision Science (AREA)
- Environmental & Geological Engineering (AREA)
- Collating Specific Patterns (AREA)
Abstract
The present invention provides a kind of intelligent advertisement delivering method based on access control system, including:1, photographic device tracking enters the face information in imaging area;2, photographic device shoots realtime graphic, the gender of analysis and identification visitor;3, the realtime graphic based on step 2 shooting, the age of analysis and identification visitor;4, according to the information for the visitor of step 2 and step 3 identified, modeling analysis is carried out, the character features of character image is confirmed, filters out the advertisement for meeting visitor in time, and launch by the display screen on gate inhibition;The present invention provides a kind of intelligent advertisement delivering method based on access control system, it can be during visitor waits for and entering, targetedly deliver advertisement, make full use of the time of waiting, can be to avoid the waste of paper resource, and the present invention can be directed to the people visitor of different sexes and age, targetedly launch advertisement, the efficiency that advertising resource is launched is improved, the benefit and value of advertisement are improved.
Description
Technical field
The present invention relates to advertisements to launch technical field, and in particular to a kind of intelligent advertisement delivery side based on access control system
Method.
Background technology
With the continuous development of society, resident family increasingly payes attention to the problem of safety, and then intelligent entrance guard is applied and given birth to,
Intelligent entrance guard is new-modernization safety management system, and it is one that it, which collects microcomputer automatic identification technology and modern safety management measure,
Body, it is related to electronics, machinery, optics, computer technology, mechanics of communication, many new technologies such as biotechnology, it is that solution is important
Department's entrance realizes the effective measures of safety precaution management;Various confidential departments are applicable in, such as bank, hotel, computer room, ordnance
Library, safe care registry, cubicle, intellectual communityintellectualized village, factory etc., in intelligent entrance guard skill today of digital technology network technology rapid development
Art has obtained swift and violent development, and intelligent access control system has surmounted simple gateway and key management already, it gradually develops
As the access management system of complete set, it plays huge effect in work circumstances safe work;In the prior art
Intelligent entrance guard visitor when visiting, generally require to wait for a period of time, in order to play advertisement function, be puted up on some doors
Some advertisements, these advertisements are single can not different classes of advertisement to be targetedly delivered to different sexes, age
The even crowd of hobby, in addition, the advertisement waste paper resource puted up, and rainwater etc. it is easy it is equal damage paper, some property are
The fine living environment of guarantee, can clear up the advertisement puted up, this results in making visitor's not can completely viewing advertisement
Or it can't see advertisement;This single inefficient blindness of advertisement delivering mode, be easy to cause the waste of advertising resource;It will also result in
The waste of paper resource does not have the effect of environmental protection;Accordingly, it is desirable to provide a kind of new technical solution is asked to solve above-mentioned technology
Topic.
Invention content
In order to overcome above-mentioned defect existing in the prior art, the present invention provides a kind of, and the intelligence based on access control system is wide
Delivering method is accused, can be to avoid the waste of paper resource, and the present invention can be directed to the people visitor of different sexes and age, there is needle
Dispensing advertisement to property improves the efficiency that advertising resource is launched, improves the benefit and value of advertisement.
The present invention improves a kind of intelligent advertisement delivering method based on access control system, including:
Step 1:Photographic device tracking mounted on a door enters the face information in imaging area;
Step 2:The photographic device shoots realtime graphic, carries out feature extraction to facial image, referred to as facial image is special
Sign vector, forms sample set, the gender of analysis and identification visitor;
Step 3:The photographic device shoots realtime graphic, carries out feature extraction to facial image, referred to as facial image is special
Sign vector, forms sample set, the age of analysis and identification visitor;
Step 4:According to the information for the visitor of step 2 and step 3 identified, modeling analysis is carried out, confirms figure map
The character features of picture filter out the advertisement for meeting visitor, and are launched by the display screen on gate inhibition in time.
The specific implementation step of the step 2 includes:Photographic device shoot realtime graphic, based on face gender algorithm from
Human face region being extracted in realtime graphic, for identification the gender of visitor, the human face region size is the dimensions of M × N,
And containing the specification of two interocular distances, ranks divide equally the face accordingly, generate grid, obtain the mesh point of matching number;It is based on
Each grid nodes extraction face subcharacter, utilizes each subcharacter information and the men and women's information being known in advance, Applied Learning algorithm
Learnt, exports training result;
The method of extraction face subcharacter is to intercept the predetermined neighborhood of corresponding mesh point first, form M1 × N1
Region, and then obtain the vector of M1 × N1 row;The value range of M1, N1 are [10,15];The gender recognition result is y=
{ 0,1 }, wherein 0 represents female, 1 represents man;
The Meshing Method is wide m deciles, and high n deciles, wherein m, n are natural number, and m ∈ [4,10], n ∈
[3,8].
The specific implementation step of the step 3 includes:Based on the realtime graphic of step 2 shooting, known based on the face age
Other algorithm extracts human face region from the realtime graphic, for identification the age of visitor, and human face region input is true in advance
Fixed age identification model extracts a n dimensional feature vector Y [y0,y1,…,yi,…,yn], wherein i ∈ [0, n], 0≤n
≤100;And by described eigenvector Y [y0,y1,…,yi,…,yn]] input age identification formula, it identifies in the realtime graphic
The age of face, wherein the age identifies that formula is:Age=arg_max (Y).
Further include recognition of face parameter, the face in the human face region extracted in the step 3 specific implementation step
Identification parameter includes eyes spacing E, cheekbone spacing B, bridge of the nose length N, forehead width F and chin width C, and the facial image is special
Sign vector is calculated as (E, B, N, F, C).
The predetermined age identification model includes training step, and the training step includes:
A, the face sample image for preparing corresponding preset quantity of each age, carries out at standardization face sample image
Reason forms age sample image;
B, it is that every age sample image marks corresponding age label, forms sample set;
C, it using the age sample image in convolutional neural networks random read take sample set, is carried from the age sample image
The corresponding feature of all ages and classes is taken, and combines this feature and generates the corresponding n dimensional feature vectors of the age sample image;
D, the penalty values for calculating the n dimensional feature vectors, utilize stochastic gradient descent method and the n-dimensional vector corresponding year
Age label is updated the parameter of the convolutional neural networks;
E, repeatedly execute step A-D, until from the penalty values of the n dimensional feature vectors extracted in age sample image no longer under
It is reduced to only.
Include step (1) between the step 3 and step 4, at the age of the visitor based on step 3 identification, carries out the age
The correction of the correction of disturbing factor parameter, the age disturbing factor includes:
Illumination detection is carried out to the realtime graphic in step 3, obtains the correction parameter of illumination parameter;
Expression Recognition is carried out to the realtime graphic in step 3, obtains the correction parameter of Expression Recognition;
The age and disturbing factor parameter based on the visitor in step 3, ageadjustment value can be obtained;The year
The sum of age of age corrected value and visitor is the actual age value of visitor;
Function between the ageadjustment value and disturbing factor parameter correction values is:
H (x)=θ0+θ1X1+θ2X2+…+θnXn
Wherein, h (x) is ageadjustment value, X1、X2、…XnIt is the correction parameter values of age disturbing factor parameter, θ respectively1、
θ2…θnIt is coefficient corresponding with the age disturbing factor correction parameter, θ0It is the correction parameter pair of institute's has age disturbing factor
The offset of the actual age value;
The range of ageadjustment value can be between [- 100,100].
Expression in the Expression Recognition includes:It is happy, sad, surprised, angry, normal.
Include step (2) between the step 3 and step 4, the gender of visitor is verified, step (2) packet
Include following steps
Step Sa:The sample set acquired in step 3 is decomposed according to age information, is divided into every subclass,
Step Sb:According to M3The decomposition of network and combined method are trained these subclass, are then combined into M3Network
Grader
Step Sc:Gender identification is carried out, is identified into comparison with the gender in step 2;
The Sa items subclass is respectively:
(E1, E2, E3, En)
(B1, B2, B3, Bn)
(N1, N2, N3, Nn)
(F1, F2, F3, Fn)
(C1, C2, C3, Cn);
The mode that the step Sb subclass is trained is to carry out linear regression processing, processing side to these subset datas
Method is:
1) x, the average value b of y are first asked;
2) equations are used:A=y-bx;
3) the formula y=bx+a equations of linear regression y=bx+a for finding out parameters crosses fixed point
(x is the age of corresponding extracting parameter personnel, and y is every subset).
The method that the step Sc carries out gender identification is:
1) data of the parameters of unknown gender personnel are extracted, including:It is long to measure eyes spacing, cheekbone spacing, the bridge of the nose
Degree, forehead width and chin width;
2) formula that the age of the personnel and parameters bring into described in corresponding parameter is fitted, carries out male respectively
Parameter fitting and women parameter fitting;
3) compare two kinds of degrees of fitting as a result, then degree of fitting it is higher be the personnel gender;
4) result obtained is compared with the gender identified in step 2, when result is consistent, is started in next step
Work, that is, carry out advertisement and delete choosing and push, when result is inconsistent, restart to detect.
Further include before the step 4:
The ad data that gender matches in character features described in typing;
The ad data that the age matches in character features described in typing.
By adopting the above-described technical solution, compared with prior art, the present invention provides a kind of based on access control system
Intelligent advertisement delivering method can targetedly be delivered advertisement, make full use of waiting during visitor waits for and entering
Time, can be to avoid the waste of paper resource, and the present invention can be directed to the people visitor of different sexes and age, targetedly
Advertisement is launched, the efficiency that advertising resource is launched is improved, improves the benefit and value of advertisement.
Description of the drawings
Fig. 1 is the flow diagram of the embodiment of the present invention 1;
Fig. 2 is the flow diagram of the embodiment of the present invention 2;
Fig. 3 is the flow diagram of the embodiment of the present invention 3;
Specific implementation mode
The specific implementation mode of the present invention is described in detail below in conjunction with attached drawing.It should be noted that this place is retouched
The specific implementation mode stated is merely to illustrate and explain the present invention, but the present invention can be defined by the claims and cover it is more
Kind different modes are implemented, and are not intended to restrict the invention.
Advertisement dispensing includes meeting-place showcase, fixation of advertisement platform, large screen rolling advertisement, advertisement dispensing under prior art center line
Unilaterally to launch, the direct interaction of shortage and user at all can not be according to user characteristics, such as:Age, gender carry out accurate wide
It accuses and launches;Especially on gate inhibition, it is manufactured almost exclusively by the prior art and puts up the mode of propagating poster and publicized, propagating poster
It is easily damaged by extraneous factor, the factors such as Ru Shui, wind;And during dispensing, many people can lose interest in, waste
Paper resource.
Present invention is primarily aimed at a kind of intelligent advertisement delivering method based on access control system is provided, to realize under line extensively
The accurate dispensing accused.
Embodiment 1:
Refering to what is shown in Fig. 1, the present invention provides a kind of intelligent advertisement delivering method based on access control system, including:
Step 1:Photographic device tracking enters the face information in imaging area;
Step 2:Photographic device shoots realtime graphic, the gender of analysis and identification visitor;
Step 3:Photographic device shoots realtime graphic, the age of analysis and identification visitor;
Step 4:According to the information for the visitor of step 2 and step 3 identified, modeling analysis is carried out, confirms figure map
The character features of picture filter out the advertisement for meeting visitor, and are launched by the display screen on gate inhibition in time.
The intelligent advertisement put-on method can solve to apply is directed to different characteristic crowd under access control equipment center line in scene
Accurate advertisement launch problem;Technical way is to identify character features, and divide crowd using face recognition technology
Class is targetedly launched by recognition result and plays different advertisements.
The gender of analysis and identification visitor and age are carried out by other processing units in step 2 and step 3, no
It is handled by photographic device.
Step 1 includes:The face information is one in whole face or selection eyes, eyebrow, nose, face
Or it is multiple.
The specific implementation step of the step 2 includes:Photographic device shoots realtime graphic, and gender identification device is based on people
Face gender algorithm extracts human face region from realtime graphic, for identification the gender of visitor;The human face region extracted can be with
It is whole face, can also be the key features such as eyes, eyebrow, nose, the face of face, when selecting local feature, to the greatest extent may be used
The position that gender discrimination can be selected larger can match a variety of positions and distinguish when discrimination is relatively small, to meet
The needs of information fusion;When the human face region extracted is whole face, it is preferred that peak width 100-150, region are high
Degree is 150-200, and all information of face is more adequately utilized in the method;When the human face region extracted is eyebrow or eye
When the region of eyeball, it is preferred that the size of image is 100 × 40, and the regional area of face is utilized in the method, but due to eyebrow eye
Eyeball has good discrimination, can also have good recognition effect;
The human face region size is the dimensions of M × N, and containing the specification of two interocular distances, and ranks are divided equally accordingly
The face generates grid, obtains the mesh point of matching number;Equidistant division net is distinguished to the height and width of M × n-quadrant of selection
Lattice obtain several mesh points;The Meshing Method is wide m deciles, and high n deciles, wherein m, n are natural number, and m is general
Take 1/10-the 1/4 of M;By largely testing, m is excessive to be then easy to include excessive invalid information, causes discrimination to decline, m
It is too small to be then easy to omit key message, equally discrimination is caused to decline, in this section, preferable discrimination can be obtained;N mono-
As take 1/8-the 1/3 of N, by verification, n is excessive or too small also all discrimination will be made to have different degrees of decline;It is preferred, therefore, that
M ∈ [4,10], n ∈ [3,8];Based on each grid nodes extraction face subcharacter, choosing 2K, (K is natural number, generally takes 100-
300, can get has preferable representative data) the normalized facial image of size, men and women's image each K, base
In each grid nodes extraction face subcharacter learning algorithm is used using each subcharacter information and the men and women's information being known in advance
Learnt;The method of extraction face subcharacter is to intercept the predetermined neighborhood of corresponding mesh point first, form M1 × N1
Region, and then obtain the vector of M1 × N1 row;M1, N1 are natural number, and the value range of M1, N1 are [10,15];The property
Other recognition result is y={ 0,1 }, wherein 0 represents female, 1 represents man;Such method is simple and clear, easily operated;M1, N1
Between taking 10-15, it is excessive or it is too small can all learning effect be caused to be deteriorated, by verification in desired discrimination above range
The peak value that can be used, and can go in the above range;It should be appreciated that the range including M1, N1 value range substantially meets one
Determine the curve of parameter effect, thus the poor numerical value of extended effect, belong to its simple transformation, it should be fallen into and protect model
Within enclosing.
The specific implementation step of the step 3 includes:The realtime graphic of photographic device shooting, age identification device are based on
Face age recognizer extracts human face region from the realtime graphic, for identification the age of visitor, by the human face region
Predetermined age identification model is inputted, a n dimensional feature vector Y [y is extracted0,y1,…,yi,…,yn], wherein i ∈
[0, n], 0≤n≤100;And by described eigenvector Y [y0,y1,…,yi,…,yn]] the input age identifies formula, described in identification
The age of face in realtime graphic, wherein the age identifies that formula is:Age=arg_max (Y).
Further include recognition of face parameter, the face in the human face region extracted in the step 3 specific implementation step
Identification parameter includes eyes spacing E, cheekbone spacing B, bridge of the nose length N, forehead width F and chin width C, and the facial image is special
Sign vector is calculated as (E, B, N, F, C);These information are the more representative information of face, can help effectively to judge personnel's
Gender and age, the present invention classify these facial informations according to the age, identical parameter are formed a set, to visiting
The gender of visitor is verified, and the accuracy of sex-screening is further improved.
The predetermined age identification model includes training step, and the training step includes:
A, the face sample image for preparing corresponding preset quantity of each age, carries out at standardization face sample image
Reason forms age sample image;
B, it is that every age sample image marks corresponding age label, forms sample set;
C, it using the age sample image in convolutional neural networks random read take sample set, is carried from the age sample image
The corresponding feature of all ages and classes is taken, and combines this feature and generates the corresponding n dimensional feature vectors of the age sample image;
D, the penalty values for calculating the n dimensional feature vectors, utilize stochastic gradient descent method and the n-dimensional vector corresponding year
Age label is updated the parameter of the convolutional neural networks;
E, repeatedly execute step A-D, until from the penalty values of the n dimensional feature vectors extracted in age sample image no longer under
It is reduced to only.
To identify in realtime graphic for the age of face, the concrete scheme of the present invention is illustrated.Work as photographic device
A realtime graphic is taken, age identification device obtains the region of face and preserves, and this completes a faces
The process of extracted region.In other embodiments, the face recognition algorithms for human face region being extracted from the realtime graphic can be with
For:Method, Local Features Analysis method, eigenface method, the method based on elastic model, neural network based on geometric properties
Method etc..
The human face region extracted using face recognition algorithms is saved as to the picture of default size, for example, saving as 256*
Picture P comprising human face region is inputted predetermined age identification model, to be carried from picture P by the picture P of 256 pixels
The middle face characteristic for representing all ages and classes is taken, the feature vector Y [y that the face characteristic generates picture P are combined0,y1,…,yi,…,
yn], wherein the predetermined age identification model is obtained by training convolutional neural networks, in the present embodiment, convolution
Neural network is inception-resnet networks, and specific training step includes:
Prepare the face sample image of corresponding preset quantity respectively for each age, for example, prepare 0 years old to 100 years old it is right
The face sample image answered forms age sample image, according to every age after carrying out standardization processing to these sample images
The age of face in sample image is that every age sample image marks age label " 0 " to " 100 ", all age samples,
Image and its age label form sample set;Convolutional neural networks are initialized, keep the feature vector of its subsequent extracted equal
For n (for example, 101) dimensions, during using sample set training convolutional neural networks, convolutional neural networks are from sample set
Random read take age sample image extracts the face characteristic of reflection all ages and classes, group from the age sample image of reading
It closes the face characteristic and generates the corresponding n dimensional feature vectors of the age sample image, often extract m (for example, 100) Zhang Nianling sample graphs
After the feature vector of picture, the penalty values (i.e. Loss) of m (for example, 100) a dimensional feature vector are calculated.Specifically, the Loss
Calculation formula it is as follows:
Wherein, XiIndicate the feature vector Y, C at ageyiIndicate the central feature vector as feature vector Y dimensions, i.e.,
YiThe eigencenter of class, and CyiInitialization value be full 0, W indicates the parameter square of the full articulamentum of the convolutional neural networks
Battle array, b indicate biasing, wjIndicate (in the present embodiment, the jth row of W, m indicate the number of samples of update model parameter input
100) m is.By calculating the Loss of feature vector, stochastic gradient descent method and the corresponding age label of described eigenvector are utilized
The parameter of the convolutional neural networks is updated, the feature vector of extraction is made more to cluster, is also made subsequently from realtime graphic
The more approaching to reality age at the age identified.The method that model parameter is updated using stochastic gradient descent method is more ripe, here
It repeats no more.Step A-D is executed repeatedly, until the Loss for the feature vector extracted from age sample image no longer declines, is stopped
Only model parameter updates, that is to say, that model training process terminates, and has obtained the Model Identification model.
It should be noted that described include to face sample image progress standardization processing:The sample that first sample is concentrated
This picture is pre-processed such as scaling, cutting, overturning and/or distortion operation, utilizes the face sample after standardization processing
This image is trained convolutional neural networks, effectively improves the authenticity and accuracy rate of model training.
Further include before the step 4:
The ad data that gender matches in character features described in typing;
The ad data that the age matches in character features described in typing.
Preferably, in the network terminal typing with the ad data that gender matches in the character features and with it is described
The ad data that the age matches in character features;By wireless network module, quick information friendship can be carried out with the network terminal
It changes, filters out the advertisement type for meeting consumer in time, the network terminal includes consumer's big data module or advertisement big data mould
Block actively acquires the information of visitor by photographic device, and the target for meeting consumer type is filtered out from the network terminal, further according to
Target intelligently selects advertisement type from the network terminal, and advertisement is thrown and is played on delivery device, and the delivery device is preferentially adopted
With the display screen being arranged on entrance guard device, the present invention can intelligent recognition consumer type, targetedly advertisement is discriminated
Choosing is launched, and advertisement benefit and value are greatly improved.
Embodiment 2:
Refering to what is shown in Fig. 2, the place different from embodiment 1 is, include step S between the step 3 and step 4
(1):At the age of visitor based on step 3 identification, the correction of age disturbing factor parameter is carried out, the age disturbing factor
Correction includes:
Illumination detection is carried out to the realtime graphic in step 3, obtains the correction parameter of illumination parameter;
Expression Recognition is carried out to the realtime graphic in step 3, obtains the correction parameter of expression parameter;
The expression parameter includes:It is happy, sad, surprised, angry, normal.
Optionally, the age disturbing factor parameter includes at least one in illumination parameter, expression parameter.
By taking illumination parameter as an example, if stronger relative to same target face ambient light photograph, i.e., the described target facial image
In light intensity value it is big, then carrying out the age value that obtains of age identification can be smaller than actual age value;, whereas if ambient light is shone
Weaker, i.e., the light intensity value in the described target facial image is small, then carrying out the age value that obtains of age identification can be than practical year
Age value is big.
For human face expression parameter, human face expression can be indignation, sadness, happiness etc..For example, relative to same
If the human face expression of the target face target face is happiness, carrying out the age value that obtains of age identification can be than practical year
Age value is small;If the human face expression of the target face is sadness, carrying out the age value that obtains of age identification can be than practical year
Age value is big.
Certainly, for the face ageadjustment parameter other than the above parameter, can also include other can influence age identification
As a result parameter, such as human face posture, face look up angle difference, can also influence the size of age value, the embodiment of the present disclosure
Seek to determine the correction parameter values of different face interference parameters to ageadjustment value, to improve the accurate of age estimation
Property.
It, can be to all institutes as training data after establishing the face age models in the embodiment of the present disclosure
It states facial image and is based on ageadjustment parameter progress statistical test, obtain the ageadjustment parameter and the face age is identified
Influence degree, it is alternatively possible to describe the influence degree with ageadjustment value.
Assuming that only include illumination parameter in the face ageadjustment parameter, then it can be to all institutes as training data
It states facial image and is based on the illumination parameter progress statistical test, illumination inspection is carried out to the facial image according to the relevant technologies
It surveys, tests out different illumination parameter values, and determine according to the illumination parameter value tested out and carried out to the facial image
The ageadjustment value of age identification;It can determine that all facial images are each according to the face age models having built up
From estimate age value, the respective actual age value of all facial images is had determined that before acquiring training data,
By the actual age value of each facial image subtract it is described estimate age value, just obtained the illumination parameter to institute
State the correction age value of facial image.
Such as illumination parameter value is a1When, facial image A is b by the age value that step 3 determines1, acquire training data
When the actual age value of the facial image A that has determined be b2, then illumination parameter value is a1When, carrying out age identification can make
At ageadjustment value be (b2-b1)。
Age based on the visitor in step 3 and disturbing factor parameter, can obtain ageadjustment value;The age school
The sum of age of positive value and visitor is the actual age value of visitor;
Function between the ageadjustment value and disturbing factor parameter correction values is:
H (x)=θ0+θ1X1+θ2X2+…+θnXn;
Wherein, h (x) is ageadjustment value, X1、X2、…XnIt is the correction parameter values of age disturbing factor parameter, θ respectively1、
θ2…θnIt is coefficient corresponding with the age disturbing factor correction parameter, θ0It is the correction parameter pair of institute's has age disturbing factor
The offset of the actual age value.
The range of ageadjustment value can be between [- 100,100].
For example, X1The illumination parameter value of the illumination parameter in face ageadjustment parameter, X can be represented2Face can be represented
The human face expression parameter value of human face expression parameter in ageadjustment parameter, XnFace age interference parameter correction ginseng can be represented
The corresponding parameter value of n-th of parameter in number;θ1、θ2…θnIllumination parameter, expression parameter and n-th of parameter are corresponded to respectively, and
θ0、θ1、θ2…θnAfter once it is determined that, it will not change relative to same face age numerical value;
The face age models are established by the above process, and after the estimation function is determined, based on same
When face age models carry out age identification, it is no longer necessary to repeat and establish faceform and acquisition target facial image;?
When needing to carry out age identification, step 3 is executed, obtains the image of visitor's face.
In step 3, the image of visitor's face can intercept the individual picture shot from photographic device, also may be used
To intercept from a certain frame picture in video;It can be according to the relevant technologies, to the advanced row recognition of face of picture, such as in video
Each frame picture carry out recognition of face, if it includes face to recognize a certain picture, it includes people to intercept the picture
The image of face is as visitor's facial image.
Determine that the target light of illumination parameter is as follows according to the process of parameter value:It can be to visitor's facial image according to correlation
Technology carries out illumination detection, to obtain the target light of the illumination parameter according in parameter value, such as the target facial image
Light intensity value etc..
Determine that the process of visitor's mark expression parameter value of human face expression parameter is as follows:Visitor's facial image can be pressed
Facial expression recognition is carried out according to the relevant technologies, obtains the target human face expression parameter value of the human face expression parameter.
Determine that the process of visitor's human face posture parameter value of human face posture parameter is as follows:It can be to the target facial image
Human face posture detection is carried out according to the relevant technologies, obtains the target face attitude parameter value of the human face posture parameter, such as visit
Rotation angle value, pitch angle angle value of guest's face etc..
The sum of age of the ageadjustment value and visitor is the actual age value of visitor;By the age interference parameter
Correction parameter values substitute into the estimation function, and the ageadjustment value is calculated.Wherein, the target correction age value
Range is equally between [- 100,100];The sum of the age value of the visitor detected in the step 3 and the ageadjustment value are true
It is set to visitor's actual age value.
In this step, the sum of age value and ageadjustment value of the visitor detected in the step 3 can be directly calculated,
The result of calculating is finally determined as visitor's actual age value.
In above-mentioned steps, the age value of visitor is first determined, while obtaining age disturbing factor correction parameter in the visitor
Ageadjustment value in facial image, and then according to the age value of the visitor and the ageadjustment value, determine the visitor
Actual age value;The age is known due to considering age interference parameter correction parameter in above-mentioned face age identification process
Other influence, therefore improve the accuracy of face age identification.
Embodiment 3
With reference to figure 3, the place different from embodiment 2 is, is used for including step (2) between the step 3 and step 4
The gender of visitor is verified, keeps the testing result to visitor's gender more accurate, the step (2) includes the following steps
Step Sa:The sample set acquired in step 3 is decomposed according to age information, is divided into every subclass,
Step Sb:According to M3The decomposition of network and combined method are trained these subclass, are then combined into M3Network
Grader
Step Sc:Gender identification is carried out, is identified into comparison with the gender in step 2;
The Sa items subclass is respectively:
(E1, E2, E3, En)
(B1, B2, B3, Bn)
(N1, N2, N3, Nn)
(F1, F2, F3, Fn)
(C1, C2, C3, Cn).
The mode that the step Sb subclass is trained is to carry out linear regression processing, processing side to these subset datas
Method is:
1) x, the average value b of y are first asked;
2) equations are used:A=y-bx;
3) the formula y=bx+a equations of linear regression y=bx+a for finding out parameters crosses fixed point
(x is the age of corresponding extracting parameter personnel, and y is every subset).
Step Sc carries out gender identification, and method is:
1) data of the parameters of unknown gender personnel are extracted, including:It is long to measure eyes spacing, cheekbone spacing, the bridge of the nose
Degree, forehead width and chin width;
2) formula that the age of the personnel and parameters bring into described in corresponding parameter is fitted, carries out male respectively
Parameter fitting and women parameter fitting;
3) compare two kinds of degrees of fitting as a result, then degree of fitting it is higher be the personnel gender;
4) result obtained is compared with the gender identified in step 2, when result is consistent, is started in next step
Work, that is, carry out advertisement and delete choosing and push, when result is inconsistent, restart to detect;
Preferred in the present invention, photographic device uses camera, and the information of visitor is acquired using face recognition technology, utilizes
Processing system, the gender of analysis and identification visitor and age, such as PLC controller, processing system are connect with by message exchange,
Message exchange connect connection with the network terminal, the target for meeting consumer type is filtered out by the network terminal, further according to mesh
Mark from the network terminal intelligently selection advertisement type, and by advertisement throwing play on a display screen, the display screen it is mounted on a door or
Person is mounted on the both sides of door, the place for facilitating visitor to be watched;Described information exchanger preferentially uses wireless network module.
In conclusion by adopting the above-described technical solution, compared with prior art, the present invention provides one kind being based on door
The intelligent advertisement delivering method of access control system can targetedly deliver advertisement, fully profit during visitor waits for and entering
, can be to avoid the waste of paper resource with the time of waiting, and the present invention can be directed to the people visitor of different sexes and age, have
Advertisement is targetedly launched, the efficiency that advertising resource is launched is improved, improves the benefit and value of advertisement.
The preferred embodiment of the present invention is described in detail above in association with attached drawing, still, the present invention is not limited to above-mentioned embodiment party
Detail in formula can carry out a variety of simple changes to technical scheme of the present invention within the scope of the technical concept of the present invention
Type, these simple variants all belong to the scope of protection of the present invention.
It is further to note that specific technical features described in the above specific embodiments, in not lance
In the case of shield, it can be combined by any suitable means, in order to avoid unnecessary repetition, the various possibility of the present invention
Combination no longer separately illustrate.
In addition, various embodiments of the present invention can be combined randomly, as long as it is without prejudice to originally
Invention thought, it should also be regarded as the disclosure of the present invention.
Claims (9)
1. the present invention improves a kind of intelligent advertisement delivering method based on access control system, including:
Step 1:Photographic device tracking mounted on a door enters the face information in imaging area;
Step 2:The photographic device shoots realtime graphic, and feature extraction is carried out to facial image, referred to as facial image feature to
Amount forms sample set, the gender of analysis and identification visitor;
Step 3:The photographic device shoots realtime graphic, and feature extraction is carried out to facial image, referred to as facial image feature to
Amount forms sample set, the age of analysis and identification visitor;
Step 4:According to the information for the visitor of step 2 and step 3 identified, modeling analysis is carried out, confirms character image
Character features filter out the advertisement for meeting visitor, and are launched by the display screen on gate inhibition in time.
2. a kind of intelligent advertisement delivering method based on access control system according to claim 1, which is characterized in that the step
Rapid two specific implementation step includes:Photographic device shoots realtime graphic, is extracted from realtime graphic based on face gender algorithm
Go out human face region, for identification the gender of visitor, the human face region size is the dimensions of M × N, and containing between two pupils
Away from specification, accordingly ranks divide equally the face, generate grid, obtain the mesh point of matching number;Based on each grid nodes extraction
Face subcharacter, using each subcharacter information and the men and women's information being known in advance, Applied Learning algorithm is learnt, output instruction
Practice result;The method of extraction face subcharacter is to intercept the predetermined neighborhood of corresponding mesh point first, form M1 × N1 sub-district
Domain, and then obtain the vector of M1 × N1 row;The value range of M1, N1 are [10,15];The gender recognition result be y=0,
1 }, wherein 0 represents female, 1 represents man;
The Meshing Method is wide m deciles, and high n deciles, wherein m, n are natural number, and m ∈ [4,10], n ∈ [3,8].
3. a kind of intelligent advertisement delivering method based on access control system according to claim 1, which is characterized in that the step
Rapid three specific implementation step includes:It is real-time from this based on face age recognizer based on the realtime graphic of step 2 shooting
Human face region is extracted in image, for identification the age of visitor, which is inputted into predetermined age identification mould
Type extracts a n dimensional feature vector Y [y0,y1,…,yi,…,yn], wherein i ∈ [0, n], 0≤n≤100;And it will be described
Feature vector Y [y0,y1,…,yi,…,yn]] input age identification formula, identify the age of face in the realtime graphic,
In, the age identification formula is:Age=arg_max (Y).
4. a kind of intelligent advertisement delivering method based on access control system according to claim 3, which is characterized in that described to carry
Further include recognition of face parameter in the human face region taken, the recognition of face parameter includes eyes spacing E, cheekbone spacing B, the bridge of the nose
Length N, forehead width F and chin width C, the facial image feature vector are calculated as (E, B, N, F, C).
5. a kind of intelligent advertisement delivering method based on access control system according to claim 3, which is characterized in that described pre-
First determining age identification model includes training step, and the training step includes:
A, the face sample image for preparing corresponding preset quantity of each age carries out standardization processing shape to face sample image
At age sample image;
B, it is that every age sample image marks corresponding age label, forms sample set;
C, it using the age sample image in convolutional neural networks random read take sample set, is extracted not from the age sample image
The corresponding feature with the age, and combine this feature and generate the corresponding n dimensional feature vectors of the age sample image;
D, the penalty values for calculating the n dimensional feature vectors are marked using stochastic gradient descent method and the n-dimensional vector corresponding age
Label are updated the parameter of the convolutional neural networks;
E, step A-D is executed repeatedly, until the penalty values of the n dimensional feature vectors extracted from age sample image are no longer fallen to
Only.
6. a kind of intelligent advertisement delivering method based on access control system according to claim 1, which is characterized in that the step
Include step (1) between rapid three and step 4, at the age of the visitor based on step 3 identification, carries out age disturbing factor parameter
Correction, the correction of the age disturbing factor include:
Illumination detection is carried out to the realtime graphic in step 3, obtains the correction parameter of illumination parameter;
Expression Recognition is carried out to the realtime graphic in step 3, obtains the correction parameter of Expression Recognition;
The age and disturbing factor parameter based on the visitor in step 3, ageadjustment value can be obtained;The age school
The sum of age of positive value and visitor is the actual age value of visitor;
Function between the ageadjustment value and disturbing factor parameter correction values is:
H (x)=θ0+θ1X1+θ2X2+…+θnXn
Wherein, h (x) is ageadjustment value, X1、X2、…XnIt is the correction parameter values of age disturbing factor parameter, θ respectively1、θ2…
θnIt is coefficient corresponding with the age disturbing factor correction parameter, θ0It is the correction parameter of institute's has age disturbing factor to described
The offset of actual age value;
The range of ageadjustment value can be between [- 100,100].
7. a kind of intelligent advertisement delivering method based on access control system according to claim 6, which is characterized in that the table
Feelings identification in expression include:It is happy, sad, surprised, angry, normal.
8. a kind of intelligent advertisement delivering method based on access control system according to claim 1, which is characterized in that the step
Include step (2) between rapid three and step 4, the gender of visitor is verified, the step (2) includes the following steps:
Step Sa:The sample set acquired in step 3 is decomposed according to age information, is divided into every subclass;
The Sa items subclass is respectively:
(E1, E2, E3, En)
(B1, B2, B3, Bn)
(N1, N2, N3, Nn)
(F1, F2, F3, Fn)
(C1, C2, C3, Cn);
Step Sb:According to M3The decomposition of network and combined method are trained these subclass, are then combined into M3Network class
Device;
The mode that the step Sb subclass is trained is to carry out linear regression processing, processing method to these subset datas
For:
1) x, the average value b of y are first asked;
2) equations are used:A=y-bx;
3) the formula y=bx+a equations of linear regression y=bx+a for finding out parameters crosses fixed point
(x is the age of corresponding extracting parameter personnel, and y is every subset);
Step Sc:Gender identification is carried out, is identified into comparison with the gender in step 2;
The method that the step Sc carries out gender identification is:
1) data of the parameters of unknown gender personnel are extracted, including:Measure eyes spacing, cheekbone spacing, bridge of the nose length, volume
Head width and chin width;
2) formula that the age of the personnel and parameters are brought into corresponding parameter is fitted, and carries out male's parameter fitting respectively
With women parameter fitting;
3) compare two kinds of degrees of fitting as a result, then degree of fitting it is higher be the personnel gender;
4) result obtained is compared with the gender identified in step 2, when result is consistent, starts the work of next step
Make, that is, carry out advertisement deletes choosing and push, when result is inconsistent, restarts to detect.
9. a kind of intelligent advertisement delivery side based on access control system according to claim 1 to 8 any one claim
Method, which is characterized in that further include before the step 4:
The ad data that gender matches in character features described in typing;
The ad data that the age matches in character features described in typing.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810540510.6A CN108765014A (en) | 2018-05-30 | 2018-05-30 | A kind of intelligent advertisement put-on method based on access control system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810540510.6A CN108765014A (en) | 2018-05-30 | 2018-05-30 | A kind of intelligent advertisement put-on method based on access control system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108765014A true CN108765014A (en) | 2018-11-06 |
Family
ID=64004442
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810540510.6A Pending CN108765014A (en) | 2018-05-30 | 2018-05-30 | A kind of intelligent advertisement put-on method based on access control system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108765014A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110991213A (en) * | 2019-05-14 | 2020-04-10 | 杨春燕 | Automatic big data distribution platform |
CN110991256A (en) * | 2019-11-11 | 2020-04-10 | 无锡慧眼人工智能科技有限公司 | System and method for carrying out age estimation and/or gender identification based on face features |
CN111340523A (en) * | 2020-01-17 | 2020-06-26 | 北京鸿途信达科技股份有限公司 | Advertisement putting method and device and computer storage medium |
CN111738749A (en) * | 2019-06-18 | 2020-10-02 | 北京京东尚科信息技术有限公司 | Information display method and device, electronic equipment and storage medium |
CN112883731A (en) * | 2021-04-29 | 2021-06-01 | 腾讯科技(深圳)有限公司 | Content classification method and device |
CN113747640A (en) * | 2021-09-03 | 2021-12-03 | 深圳时空数字科技有限公司 | Intelligent central control method and system for digital exhibition hall lamplight |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103034840A (en) * | 2012-12-05 | 2013-04-10 | 山东神思电子技术股份有限公司 | Gender identification method |
US20130241817A1 (en) * | 2012-03-16 | 2013-09-19 | Hon Hai Precision Industry Co., Ltd. | Display device and method for adjusting content thereof |
CN104464056A (en) * | 2014-12-12 | 2015-03-25 | 北京中研众力科技有限公司 | Facial recognition access control equipment with advertising information media function |
CN106022317A (en) * | 2016-06-27 | 2016-10-12 | 北京小米移动软件有限公司 | Face identification method and apparatus |
CN106096580A (en) * | 2016-06-28 | 2016-11-09 | 合肥酷睿网络科技有限公司 | A kind of network by recognition of face method for distinguishing |
CN106815566A (en) * | 2016-12-29 | 2017-06-09 | 天津中科智能识别产业技术研究院有限公司 | A kind of face retrieval method based on multitask convolutional neural networks |
CN106845330A (en) * | 2016-11-17 | 2017-06-13 | 北京品恩科技股份有限公司 | A kind of training method of the two-dimension human face identification model based on depth convolutional neural networks |
CN106887062A (en) * | 2017-02-13 | 2017-06-23 | 合肥美信息科技有限公司 | Recognition of face advertisement gate inhibition's main frame based on high in the clouds |
CN107423707A (en) * | 2017-07-25 | 2017-12-01 | 深圳帕罗人工智能科技有限公司 | A kind of face Emotion identification method based under complex environment |
CN107844744A (en) * | 2017-10-09 | 2018-03-27 | 平安科技(深圳)有限公司 | With reference to the face identification method, device and storage medium of depth information |
CN107977633A (en) * | 2017-12-06 | 2018-05-01 | 平安科技(深圳)有限公司 | Age recognition methods, device and the storage medium of facial image |
-
2018
- 2018-05-30 CN CN201810540510.6A patent/CN108765014A/en active Pending
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130241817A1 (en) * | 2012-03-16 | 2013-09-19 | Hon Hai Precision Industry Co., Ltd. | Display device and method for adjusting content thereof |
CN103034840A (en) * | 2012-12-05 | 2013-04-10 | 山东神思电子技术股份有限公司 | Gender identification method |
CN104464056A (en) * | 2014-12-12 | 2015-03-25 | 北京中研众力科技有限公司 | Facial recognition access control equipment with advertising information media function |
CN106022317A (en) * | 2016-06-27 | 2016-10-12 | 北京小米移动软件有限公司 | Face identification method and apparatus |
CN106096580A (en) * | 2016-06-28 | 2016-11-09 | 合肥酷睿网络科技有限公司 | A kind of network by recognition of face method for distinguishing |
CN106845330A (en) * | 2016-11-17 | 2017-06-13 | 北京品恩科技股份有限公司 | A kind of training method of the two-dimension human face identification model based on depth convolutional neural networks |
CN106815566A (en) * | 2016-12-29 | 2017-06-09 | 天津中科智能识别产业技术研究院有限公司 | A kind of face retrieval method based on multitask convolutional neural networks |
CN106887062A (en) * | 2017-02-13 | 2017-06-23 | 合肥美信息科技有限公司 | Recognition of face advertisement gate inhibition's main frame based on high in the clouds |
CN107423707A (en) * | 2017-07-25 | 2017-12-01 | 深圳帕罗人工智能科技有限公司 | A kind of face Emotion identification method based under complex environment |
CN107844744A (en) * | 2017-10-09 | 2018-03-27 | 平安科技(深圳)有限公司 | With reference to the face identification method, device and storage medium of depth information |
CN107977633A (en) * | 2017-12-06 | 2018-05-01 | 平安科技(深圳)有限公司 | Age recognition methods, device and the storage medium of facial image |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110991213A (en) * | 2019-05-14 | 2020-04-10 | 杨春燕 | Automatic big data distribution platform |
CN111738749A (en) * | 2019-06-18 | 2020-10-02 | 北京京东尚科信息技术有限公司 | Information display method and device, electronic equipment and storage medium |
CN110991256A (en) * | 2019-11-11 | 2020-04-10 | 无锡慧眼人工智能科技有限公司 | System and method for carrying out age estimation and/or gender identification based on face features |
CN111340523A (en) * | 2020-01-17 | 2020-06-26 | 北京鸿途信达科技股份有限公司 | Advertisement putting method and device and computer storage medium |
CN112883731A (en) * | 2021-04-29 | 2021-06-01 | 腾讯科技(深圳)有限公司 | Content classification method and device |
CN112883731B (en) * | 2021-04-29 | 2021-08-20 | 腾讯科技(深圳)有限公司 | Content classification method and device |
CN113747640A (en) * | 2021-09-03 | 2021-12-03 | 深圳时空数字科技有限公司 | Intelligent central control method and system for digital exhibition hall lamplight |
CN113747640B (en) * | 2021-09-03 | 2024-02-09 | 深圳时空数字科技有限公司 | Intelligent central control method and system for digital exhibition hall lamplight |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108765014A (en) | A kind of intelligent advertisement put-on method based on access control system | |
CN111325115B (en) | Cross-modal countervailing pedestrian re-identification method and system with triple constraint loss | |
CN105426827B (en) | Living body verification method, device and system | |
CN105243386B (en) | Face living body judgment method and system | |
Raghavendra et al. | Exploring the usefulness of light field cameras for biometrics: An empirical study on face and iris recognition | |
Kashem et al. | Face recognition system based on principal component analysis (PCA) with back propagation neural networks (BPNN) | |
CN106897675A (en) | The human face in-vivo detection method that binocular vision depth characteristic is combined with appearance features | |
JP2013242825A (en) | Personal attribute estimation system and learning data creation device | |
CN108681928A (en) | A kind of intelligent advertisement put-on method | |
CN111222380B (en) | Living body detection method and device and recognition model training method thereof | |
WO2020172870A1 (en) | Method and apparatus for determining motion trajectory of target object | |
CN111178130A (en) | Face recognition method, system and readable storage medium based on deep learning | |
CN107392191B (en) | Method for judging identity, device and electronic equipment | |
CN112541421A (en) | Pedestrian reloading identification method in open space | |
CN110929570B (en) | Iris rapid positioning device and positioning method thereof | |
Haq et al. | COMSATS face: A dataset of face images with pose variations, its design, and aspects | |
Zeinstra et al. | ForenFace: a unique annotated forensic facial image dataset and toolset | |
CN114581990A (en) | Intelligent running test method and device | |
Thorupunoori et al. | Camera Based Drunks Detection Mechanism Integrated with DL (Deep Learning) | |
Di Martino et al. | Rethinking shape from shading for spoofing detection | |
CN109886084B (en) | Face authentication method based on gyroscope, electronic equipment and storage medium | |
JP2015162012A (en) | Face matching device, face matching method and program | |
CN114627534B (en) | Living body discriminating method, electronic apparatus, and storage medium | |
CN110443122A (en) | Information processing method and Related product | |
CN110084110B (en) | Near-infrared face image recognition method and device, electronic equipment and storage medium |
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: 20181106 |