CN108345851A - A method of based on recognition of face analyzing personal hobby - Google Patents
A method of based on recognition of face analyzing personal hobby Download PDFInfo
- Publication number
- CN108345851A CN108345851A CN201810107657.6A CN201810107657A CN108345851A CN 108345851 A CN108345851 A CN 108345851A CN 201810107657 A CN201810107657 A CN 201810107657A CN 108345851 A CN108345851 A CN 108345851A
- Authority
- CN
- China
- Prior art keywords
- data
- age
- face
- gender
- hobby
- 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
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/178—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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention discloses a kind of methods based on recognition of face analyzing personal hobby, corresponding age, gender in face data are extracted respectively using multiple deep learning networks, and abstract facial characteristics, it finally obtains the hobby that there is the people of similar abstract facial characteristics to have using clustering method to be distributed, to recommend hobby to user.The present invention, in the advantage of processing face data, obtains the information at corresponding with the face data age and gender with high confidence level using depth learning technology；The present invention automatically extracts feature using the mode of deep learning, and the feature extracted has stronger robustness, to greatly improve the accuracy of face recognition, and improves the precision of hobby recommendation.
Description
Technical field
The invention belongs to the technical fields of information processing, and in particular to one kind being based on recognition of face analyzing personal hobby
Method.
Background technology
At this stage, marketing is launched based on the advertisement accurately to user interest taste analysis and is widely present in e-commerce initiative
In.Traditional individual subscriber hobby collection method mostly uses artificial investigation face to face or is based on the relevant technologies such as big data,
From the online activity of user, browsing, which records, and registration is gathered materials is analyzed.More or less there is inefficient, effect in such method
The bad problem of fruit.Meanwhile with the enhancing of privacy of user protective awareness, certain customers refuse to provide related data, cause
The hobby collection mode stated is hard to work, this means the loss of potential value user for advertiser and operator.
It is therefore proposed that a kind of mode of new discriminance analysis personal interest is most important.
Invention content
The purpose of the present invention is to provide a kind of method based on recognition of face analyzing personal hobby, the present invention utilizes
Depth learning technology obtains the age corresponding with face data with high confidence level and gender in the advantage of processing face data
Information；The present invention automatically extracts feature using the mode of deep learning, and the feature extracted has stronger robust
Property, to greatly improve the accuracy of face recognition, and improve the precision of hobby recommendation.
The present invention is achieved through the following technical solutions：A kind of side based on recognition of face analyzing personal hobby
Method mainly includes the following steps that：
Step E1：Collect comprising the age, gender, hobby facial image as training data；
Step E2：Using AlexNet networks as basic network structure, trained to obtain AgeNet training with the data in step E1
Model and GenderNet training patterns；The characteristic of face is obtained using age, the corresponding DeepID networks of gender, and will
Facial characteristics vectorization；The AgeNet training patterns are used to generate the mapping relations at face data and age, can obtain not
Know the age information of the face data at age；The GenderNet training patterns are used to generate the mapping of face data and gender
Relationship can obtain the gender information of the face data of unknown gender；
Step E3：Clustering Model is called after step E2, the distribution of hobby is obtained according to cluster result, to which output pushes away
The hobby recommended；
Step E4：The facial photo that user is inputted in the model that step E1-E3 is trained, using the AgeNet in step E2
Network obtains the age information of user, obtains the gender information of user using the GenderNet networks in step E2, finally calls
The information of Clustering Model output hobby in step E3.
For the present invention using depth learning technology in the advantage of processing face data, can obtain face data has high confidence
The age of degree and gender information.In crowd's group that same age section and gender form, facial feature data can be effectively
Using in terms of interest is predicted.In total solution, it would be desirable to extract face respectively using multiple deep learning networks
Corresponding age, gender in portion's data, and abstract facial characteristics finally obtain having similar abstract face special using clustering method
The hobby distribution that the people of sign has, to recommend hobby to user.
In order to preferably realize the present invention, further, according to the data in step E1 in the step E2, use
For AlexNet networks as basic network structure, every 5 years old is an age range, is carried out to the age information in training data
One-hot is encoded, and coding result is 14 dimensional vectors；By fc8 layers in AlexNet network structures of num-output parameters modification
It is 12, while is age-layer by the name modifications of fc8, and is AgeNet by modified network naming；In Caffe environment
The middle sample training using label obtains AgeNet training patterns, and the mapping relations for generating face data and age can
Obtain the age information of the face data at unknown age.
In order to preferably realize the present invention, further, use AlexNet networks as basic network in the step E2
Gender information in training data is carried out one-hot codings, and coding result by structure, the gender data in training step E1
For 2 dimensional vectors；Fc8 layers in AlexNet network structures of num-output parameters are revised as 2, while by the name modifications of fc8
For gender-layer, and it is GenderNet by modified network naming；It is instructed using the sample of label in Caffe environment
GenderNet training patterns are got, the mapping relations for generating face data and gender can obtain the face of unknown gender
The gender information of portion's data.
In order to preferably realize the present invention, further, in the step E2 screening has identical gender in training data
Sample with same age section is one group of data, and human face data is identified using the recognition of face network of entitled DeepID,
The data of the high-rise convolutional layer in DeepID networks are extracted as facial feature data.
In order to preferably realize the present invention, further, 28 groups are screened in the step E2 has identical gender and identical
The sample of age bracket.
In order to preferably realize the present invention, further, the human face data for inputting each in the step E3 corresponds to
Facial feature data be stretched as one-dimensional vector, the one-dimensional vector that whole human face data in same group of data is generated uses
Kmeans clustering methods are clustered.
In order to preferably realize the present invention, further, in the step E1 screening retain the age be more than or equal to 10 years old and
Sample less than 80 years old.
In view of all there is most of the Internet activities certain social attribute, user can upload such as true head portrait,
Age, the data such as hobby.We are based on current existing user's face data and related personal information, utilize popular depth
Learning art and traditional unsupervised-learning algorithm are spent, a set of face information identification personal interest of capable of utilizing is devised
Model.The model can provide the relevant information of the possible hobby of user in the case where only providing user's facial image,
To auxiliary ad-vertisement precision marketing activity.It can be directed to potential new user using the model with service provider, quickly obtained emerging
Interesting taste analysis as a result, so as to be unfolded advertisement precision marketing behavior in time.
According to existing human face data and one-to-one personal information data, such as：Gender, the age, native place, hobby, closely
Phase activity etc. extracts the correlated characteristic in face data, establishes facial characteristics and associated personal information using depth learning technology
Between relationship.For new, there is high confidence level, the result of high correlation to make for the only input data of face-image, output
To input the possible hobby of target.The invention mainly includes steps：
1. the preparation of training data：Before planned network, we are firstly the need of enough face image datas are collected into, together
When ensure that each image data has corresponding gender, the relevant informations such as age and hobby.We, which screen, retains the age
>=10 years old simultaneously<80 years old samples, in order to ensure that the user of the age bracket has enough hobby data；
2. the age identifies network design and training：We use AlexNet networks as basic network structure.We opened from 10 years old
Begin, one-hot codings is carried out to the age information in training data for an age range within every 5 years old（Coding result be 14 tie up to
Amount）.The num_output parameters for changing entitled fc8 layers in AlexNet network structures are 12, while being by the name modifications of fc8
Age_layer, by being named as modified network " AgeNet "；In Caffe environment, the sample training of label is used
AgeNet.The main purpose of the network is to generate the mapping relations of face data and age, can obtain the face at unknown age
The age information of data；
3. gender identifies network design and training：We use AlexNet networks as basic network structure.Our training datas
In gender information carry out one-hot codings（Coding result is 2 dimensional vectors）.It changes fc8 layers entitled in AlexNet network structures
Num_output parameters be 2, while by the name modifications of fc8 be gender_layer, by being named as modified network "
GenderNet”；In Caffe environment, the sample training GenderNet of label is used.The main purpose of the network is generation face
The mapping relations of portion's data and gender can obtain the gender information of the face data of unknown gender；
4. facial feature extraction module design and training：We screen the sample with identical gender and same age section in training set
This is one group of data（28 groups of data altogether）, in each group of data, using the recognition of face network of entitled DeepID to face
Data are identified, and extract the data of the high-rise convolutional layer in DeepID networks as facial feature data.Each is inputted
The corresponding facial feature data of human face data is stretched as 1 dimensional vector, by all human face datas in same group of data generate 1 tie up to
Amount is clustered using clustering methods such as Kmeans, is point of the corresponding hobby of of a sort data in Statistical Clustering Analysis result
Cloth situation, cluster result should preserve, for convenience of being called in service stage.
Beneficial effects of the present invention：
1. using AlexNet networks as basic network structure, trained to obtain AgeNet training patterns with the data in step E1
With GenderNet training patterns；The characteristic of face is obtained using age, the corresponding DeepID networks of gender, and will be facial
Feature vector；Clustering Model is called after step E2, the distribution of such hobby is obtained according to cluster result, it is final defeated
Go out the hobby of recommendation；The present invention is extracted the corresponding age in face data respectively using multiple deep learning networks, gender,
Abstract facial characteristics obtains there is the people of similar abstract facial characteristics to have finally using unsupervised-learning algorithms such as clusters
Hobby is distributed, to obtain possible hobby；The present invention is using depth learning technology in the excellent of processing face data
Gesture obtains age and gender information of the face data with high confidence level；The present invention is using the mode of deep learning automatically to spy
Sign extracts, and the feature extracted has stronger robustness, to greatly improve the accuracy of face recognition, and carries
The precision that high hobby is recommended.
2. the present invention, in the advantage of processing face data, can obtain face data and be set with height using depth learning technology
The age of reliability and gender information；It, can be according to the number of facial characteristics in crowd's group that same age section and gender form
According to the hobby for effectively recommending the crowd, which can provide use in the case where only providing user's facial image
The relevant information of the possible hobby in family can utilize the model that can be directed to potential new user, quickly obtain hobby
Analysis result has preferable practicability so as to which advertisement precision marketing behavior is unfolded in time.
Description of the drawings
Fig. 1 is the flow chart of the present invention；
Fig. 2 is the training flow chart of AgeNet training patterns；
Fig. 3 is the training flow chart of GenderNet training patterns.
Specific implementation mode
Embodiment 1：
A method of based on recognition of face analyzing personal hobby, mainly include the following steps that：
Step E1：Collect comprising the age, gender, hobby facial image as training data；
Step E2：Using AlexNet networks as basic network structure, trained to obtain AgeNet training with the data in step E1
Model and GenderNet training patterns；The characteristic of face is obtained using age, the corresponding DeepID networks of gender, and will
Facial characteristics vectorization；The AgeNet training patterns are used to generate the mapping relations at face data and age, can obtain not
Know the age information of the face data at age；The GenderNet training patterns are used to generate the mapping of face data and gender
Relationship can obtain the gender information of the face data of unknown gender；
Step E3：Clustering Model is called after step E2, the distribution of such hobby is obtained according to cluster result, it is final defeated
Go out the hobby of recommendation；
Step E4：The facial photo that user is inputted in the model that step E1-E3 is trained, using the AgeNet in step E2
Network obtains the age information of user, obtains the gender information of user using the GenderNet networks in step E2, finally calls
The information of Clustering Model output hobby in step E3.
As shown in Figure 1, obtaining the facial information of user first, identification is detected by model, is trained by AgeNet
Model calculates the prediction age of user, and the gender of user is calculated by GenderNet training patterns, then utilize the age,
The corresponding DeepID networks of gender obtain facial feature data, and facial characteristics vectorization calls Clustering Model, according to cluster result
The distribution of such hobby is obtained, the hobby of recommendation is finally exported.
The present invention is extracted the corresponding age in face data respectively using multiple deep learning networks, gender, is abstracted face
Feature obtains the hobby that there is the people of similar abstract facial characteristics to have finally using unsupervised-learning algorithms such as clusters
Distribution, to obtain possible hobby；The present invention, in the advantage of processing face data, obtains face using depth learning technology
Portion's data have age and the gender information of high confidence level；The present invention automatically puies forward feature using the mode of deep learning
It takes, the feature extracted has stronger robustness, to greatly improve the accuracy of face recognition, and improves interest
Like the precision recommended.
For the present invention using depth learning technology in the advantage of processing face data, can obtain face data has high confidence
The age of degree and gender information；It, can be according to the data of facial characteristics in crowd's group that same age section and gender form
Effectively recommend the hobby of the crowd, which can provide user in the case where only providing user's facial image
The relevant information of possible hobby can utilize the model that can be directed to potential new user, quickly obtain hobby point
Analysis has preferable practicability as a result, so as to which advertisement precision marketing behavior is unfolded in time.
Embodiment 2：
The present invention is advanced optimized on the basis of embodiment 1, according to the data in step E1 in the step E2, is used
For AlexNet networks as basic network structure, every 5 years old is an age range, is carried out to the age information in training data
One-hot is encoded, and coding result is 14 dimensional vectors；By fc8 layers in AlexNet network structures of num-output parameters modification
It is 12, while is age-layer by the name modifications of fc8, and is AgeNet by modified network naming；In Caffe environment
The middle sample training using label obtains AgeNet training patterns, and the mapping relations for generating face data and age can
Obtain the age information of the face data at unknown age；
It designs and changes after data cleansing arranges as shown in Fig. 2, collecting the face figure comprising age, gender, hobby
AgeNet models media layer damage and layer name；AgeNet models are trained with the data of collection, are obtained after successive ignition
AgeNet training patterns；
AlexNet networks are used to be used as basic network structure, the gender data in training step E1 that will train in the step E2
Gender information in data carries out one-hot codings, and coding result is 2 dimensional vectors；By fc8 layers in AlexNet network structures
Num-output parameters be revised as 2, while being gender-layer by the name modifications of fc8, and modified network is ordered
Entitled GenderNet；GenderNet training patterns are obtained using the sample training of label in Caffe environment, for generating face
The mapping relations of portion's data and gender can obtain the gender information of the face data of unknown gender；
It designs and changes after data cleansing arranges as shown in figure 3, collecting the face figure comprising age, gender, hobby
GenderNet models media layer damage and layer name；GenderNet models are trained with the data of collection, after successive ignition
To GenderNet training patterns；The sample with identical gender and same age section in training data is screened in the step E2
For one group of data, human face data is identified using the recognition of face network of entitled DeepID, is extracted in DeepID networks
The data of high-rise convolutional layer are as facial feature data.
The present invention will can obtain after the face image input AgeNet training patterns of user, GenderNet training patterns
The gender at the prediction age and prediction of user；To using unsupervised-learning algorithms such as clusters, obtain that there is similar abstract face
The hobby distribution that the people of feature has, to obtain possible hobby.
For the present invention using depth learning technology in the advantage of processing face data, can obtain face data has high confidence
The age of degree and gender information；It, can be according to the data of facial characteristics in crowd's group that same age section and gender form
Effectively recommend the hobby of the crowd, which can provide user in the case where only providing user's facial image
The relevant information of possible hobby can utilize the model that can be directed to potential new user, quickly obtain hobby point
Analysis has preferable practicability as a result, so as to which advertisement precision marketing behavior is unfolded in time.
The other parts of the present embodiment are same as Example 1, and so it will not be repeated.
Embodiment 3：
The present embodiment is advanced optimized on the basis of embodiment 1 or 2, the face number for inputting each in the step E3
It is stretched as one-dimensional vector according to corresponding facial feature data, the one-dimensional vector that human face data whole in same group of data is generated
It is clustered using Kmeans clustering methods；Screening retains sample of the age more than or equal to 10 years old and less than 80 years old in the step E1
This.
For the present invention using depth learning technology in the advantage of processing face data, can obtain face data has high confidence
The age of degree and gender information；It, can be according to the data of facial characteristics in crowd's group that same age section and gender form
Effectively recommend the hobby of the crowd, which can provide user in the case where only providing user's facial image
The relevant information of possible hobby can utilize the model that can be directed to potential new user, quickly obtain hobby point
Analysis has preferable practicability as a result, so as to which advertisement precision marketing behavior is unfolded in time.
The other parts of the present embodiment are identical as above-described embodiment 1 or 2, and so it will not be repeated.
The above is only presently preferred embodiments of the present invention, not does limitation in any form to the present invention, it is every according to
According to the technical spirit of the present invention to any simple modification, equivalent variations made by above example, the protection of the present invention is each fallen within
Within the scope of.
Claims (7)
1. a kind of method based on recognition of face analyzing personal hobby, which is characterized in that mainly include the following steps that：
Step E1：Collect comprising the age, gender, hobby facial image as training data；
Step E2：Using AlexNet networks as basic network structure, trained to obtain AgeNet training with the data in step E1
Model and GenderNet training patterns；The characteristic of face is obtained using age, the corresponding DeepID networks of gender, and will
Facial characteristics vectorization；The AgeNet training patterns are used to generate the mapping relations at face data and age, can obtain not
Know the age information of the face data at age；The GenderNet training patterns are used to generate the mapping of face data and gender
Relationship can obtain the gender information of the face data of unknown gender；
Step E3：Clustering Model is called after step E2, the distribution of hobby is obtained according to cluster result, to which output pushes away
The hobby recommended；
Step E4：The facial photo that user is inputted in the model that step E1-E3 is trained, using the AgeNet in step E2
Network obtains the age information of user, obtains the gender information of user using the GenderNet networks in step E2, finally calls
The information of Clustering Model output hobby in step E3.
2. a kind of method based on recognition of face analyzing personal hobby according to claim 1, which is characterized in that institute
It states according to the data in step E1 in step E2, using AlexNet networks as basic network structure, every 5 years old is an age
Section carries out one-hot codings to the age information in training data, and coding result is 14 dimensional vectors；By AlexNet networks
Fc8 layers of num-output parameters are revised as 12 in structure, while being age-layer by the name modifications of fc8, and will be after modification
Network naming be AgeNet；AgeNet training patterns are obtained using the sample training of label in Caffe environment, for generating
The mapping relations of face data and age can obtain the age information of the face data at unknown age.
3. a kind of method based on recognition of face analyzing personal hobby according to claim 2, which is characterized in that institute
It states and AlexNet networks is used to be used as basic network structure in step E2, the gender data in training step E1 will be in training data
Gender information carry out one-hot codings, and coding result be 2 dimensional vectors；By fc8 layers in AlexNet network structures of num-
Output parameters are revised as 2, while being gender-layer by the name modifications of fc8, and are by modified network naming
GenderNet；GenderNet training patterns are obtained using the sample training of label in Caffe environment, for generating facial number
According to the mapping relations with gender, the gender information of the face data of unknown gender can be obtained.
4. a kind of method based on recognition of face analyzing personal hobby according to claim 3, which is characterized in that institute
It is one group of data to state the sample screened in step E2 with identical gender and same age section in training data, and use is entitled
Human face data is identified in the recognition of face network of DeepID, extracts the data conduct of the high-rise convolutional layer in DeepID networks
Facial feature data.
5. a kind of method based on recognition of face analyzing personal hobby according to claim 4, which is characterized in that institute
It states and screens 28 groups of samples with identical gender and same age section in step E2.
6. a kind of method based on recognition of face analyzing personal hobby according to claim 1, which is characterized in that institute
It states in step E3 and the corresponding facial feature data of human face data that each inputs is stretched as one-dimensional vector, by same group of data
The one-dimensional vector that middle whole human face data generates is clustered using Kmeans clustering methods.
7. special according to a kind of method based on recognition of face analyzing personal hobby of claim 1-6 any one of them
Sign is, the sample for retaining the age more than or equal to 10 years old and less than 80 years old is screened in the step E1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810107657.6A CN108345851A (en) | 2018-02-02 | 2018-02-02 | A method of based on recognition of face analyzing personal hobby |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810107657.6A CN108345851A (en) | 2018-02-02 | 2018-02-02 | A method of based on recognition of face analyzing personal hobby |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108345851A true CN108345851A (en) | 2018-07-31 |
Family
ID=62959902
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810107657.6A Pending CN108345851A (en) | 2018-02-02 | 2018-02-02 | A method of based on recognition of face analyzing personal hobby |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108345851A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109377315A (en) * | 2018-10-25 | 2019-02-22 | 百度在线网络技术（北京）有限公司 | A kind of information recommendation method, device, electronic equipment and storage medium |
| CN110084174A (en) * | 2019-04-23 | 2019-08-02 | 杭州智趣智能信息技术有限公司 | A kind of face identification method, system and electronic equipment and storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102298709A (en) * | 2011-09-07 | 2011-12-28 | 江西财经大学 | Energy-saving intelligent identification digital signage fused with multiple characteristics in complicated environment |
| CN102930454A (en) * | 2012-10-07 | 2013-02-13 | 乐配(天津)科技有限公司 | Intelligent 3D (Three Dimensional) advertisement recommendation method based on multiple perception technologies |
| CN102968738A (en) * | 2012-12-06 | 2013-03-13 | 中国科学院半导体研究所 | Advertising system |
| CN103823816A (en) * | 2012-11-19 | 2014-05-28 | 上海亿动信息技术有限公司 | Device and method for information distribution based on user characteristic data recognition |
| CN105488478A (en) * | 2015-12-02 | 2016-04-13 | 深圳市商汤科技有限公司 | Face recognition system and method |
| CN106326441A (en) * | 2016-08-26 | 2017-01-11 | 乐视控股（北京）有限公司 | Information recommendation method and device |
| CN106846049A (en) * | 2017-01-06 | 2017-06-13 | 哈尔滨学院 | A kind of big data analysis method and its system that advertisement is associated based on electric business platform |
| CN106997629A (en) * | 2017-02-17 | 2017-08-01 | 北京格灵深瞳信息技术有限公司 | Access control method, apparatus and system |
-
2018
- 2018-02-02 CN CN201810107657.6A patent/CN108345851A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102298709A (en) * | 2011-09-07 | 2011-12-28 | 江西财经大学 | Energy-saving intelligent identification digital signage fused with multiple characteristics in complicated environment |
| CN102930454A (en) * | 2012-10-07 | 2013-02-13 | 乐配(天津)科技有限公司 | Intelligent 3D (Three Dimensional) advertisement recommendation method based on multiple perception technologies |
| CN103823816A (en) * | 2012-11-19 | 2014-05-28 | 上海亿动信息技术有限公司 | Device and method for information distribution based on user characteristic data recognition |
| CN102968738A (en) * | 2012-12-06 | 2013-03-13 | 中国科学院半导体研究所 | Advertising system |
| CN105488478A (en) * | 2015-12-02 | 2016-04-13 | 深圳市商汤科技有限公司 | Face recognition system and method |
| CN106326441A (en) * | 2016-08-26 | 2017-01-11 | 乐视控股（北京）有限公司 | Information recommendation method and device |
| CN106846049A (en) * | 2017-01-06 | 2017-06-13 | 哈尔滨学院 | A kind of big data analysis method and its system that advertisement is associated based on electric business platform |
| CN106997629A (en) * | 2017-02-17 | 2017-08-01 | 北京格灵深瞳信息技术有限公司 | Access control method, apparatus and system |
Non-Patent Citations (1)
| Title |
|---|
| WANGYONGXING: ""基于卷积神经网络的人脸属性识别研究"", 《道客巴巴》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109377315A (en) * | 2018-10-25 | 2019-02-22 | 百度在线网络技术（北京）有限公司 | A kind of information recommendation method, device, electronic equipment and storage medium |
| CN110084174A (en) * | 2019-04-23 | 2019-08-02 | 杭州智趣智能信息技术有限公司 | A kind of face identification method, system and electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107527318B (en) | Hair style replacement method based on generation countermeasure network model | |
| CN105426850B (en) | Associated information pushing device and method based on face recognition | |
| CN108509465A (en) | A kind of the recommendation method, apparatus and server of video data | |
| CN107368534B (en) | Method for predicting social network user attributes | |
| CN105760439B (en) | A kind of personage's cooccurrence relation map construction method based on specific behavior co-occurrence network | |
| CN106682108A (en) | Video retrieval method based on multi-modal convolutional neural network | |
| CN107577682A (en) | Users' Interests Mining and user based on social picture recommend method and system | |
| CN105740448B (en) | More microblogging timing abstract methods towards topic | |
| KR101224312B1 (en) | Friend recommendation method for SNS user, recording medium for the same, and SNS and server using the same | |
| Kim et al. | Exploiting web images for video highlight detection with triplet deep ranking | |
| CN106845513A (en) | Staff detector and method based on condition random forest | |
| CN107911643A (en) | Show the method and apparatus of scene special effect in a kind of video communication | |
| CN107943990A (en) | More video summarization methods of archetypal analysis technology based on Weight | |
| CN108345633A (en) | A kind of natural language processing method and device | |
| CN109272440A (en) | A kind of reduced graph generating method and system for combining text and picture material | |
| CN110414367A (en) | A kind of timing behavioral value method based on GAN and SSN | |
| CN109903053A (en) | A kind of anti-fraud method carrying out Activity recognition based on sensing data | |
| CN108446605B (en) | Double interbehavior recognition methods under complex background | |
| CN115131698A (en) | Video attribute determination method, device, equipment and storage medium | |
| CN115687760A (en) | User learning interest label prediction method based on graph neural network | |
| CN112559747A (en) | Event classification processing method and device, electronic equipment and storage medium | |
| CN108345851A (en) | A method of based on recognition of face analyzing personal hobby | |
| CN115862120A (en) | Separable variation self-encoder decoupled face action unit identification method and equipment | |
| CN110956157A (en) | Deep learning remote sensing image target detection method and device based on candidate frame selection | |
| CN112017162B (en) | Pathological image processing method, pathological image processing device, storage medium and processor |
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 |
Application publication date: 20180731 |
|
| RJ01 | Rejection of invention patent application after publication |