CN110490149A - A kind of face identification method and device based on svm classifier - Google Patents
A kind of face identification method and device based on svm classifier Download PDFInfo
- Publication number
- CN110490149A CN110490149A CN201910778321.7A CN201910778321A CN110490149A CN 110490149 A CN110490149 A CN 110490149A CN 201910778321 A CN201910778321 A CN 201910778321A CN 110490149 A CN110490149 A CN 110490149A
- Authority
- CN
- China
- Prior art keywords
- svm classifier
- facial characteristics
- training
- face
- obtains
- 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
- 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
- G06V40/172—Classification, e.g. identification
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
This application discloses a kind of face identification method and device based on svm classifier characterized by comprising firstly, extracting the facial characteristics that network extracts preset human face data collection by preset features, obtain the facial characteristics vector set being made of facial characteristics vector;Secondly, the facial characteristics vector set is divided into training feature vector collection and testing feature vector collection;Then the SVM classifier based on LP norm is trained with the training feature vector collection, obtains the SVM classifier of training completion;It is tested finally, the testing feature vector collection is inputted in the SVM classifier that the training is completed, obtains test result.Present application addresses the face identification methods of neural network when identifying to the facial image with Complex Noise, the poor technical problem of robustness.
Description
Technical field
This application involves technical field of face recognition more particularly to a kind of face identification methods and dress based on svm classifier
It sets.
Background technique
Recognition of face is a kind of biological identification technology for carrying out identification based on facial feature information of people, is widely answered
Need to carry out the field of authentication for safety monitoring, access control and attendance etc..
The successful key of face identification system is whether possess efficient core algorithm, and neural network is led in image recognition
Domain, which is rapidly developed, has gradually replaced traditional face recognition algorithms.Compared to traditional algorithm, neural network is to facial image end
To the processing at end, reduces the artificial work for extracting characteristics of image, accelerate identification process, it can using the method constantly learnt
Adaptive identification is carried out to the image of different quality, but currently based on the face identification method of neural network to multiple
When the facial image of miscellaneous noise is identified, robustness is poor.
Summary of the invention
This application provides a kind of face identification method and device based on svm classifier, for solving the people of neural network
Face recognition method is when identifying the facial image with Complex Noise, the poor technical problem of robustness.
In view of this, the application first aspect provides a kind of face identification method based on svm classifier, comprising:
The facial characteristics that network extracts preset human face data collection is extracted by preset features, obtains being made of facial characteristics vector
Facial characteristics vector set;
The facial characteristics vector set is divided into training feature vector collection and testing feature vector collection;
The SVM classifier based on LP norm is trained with the training feature vector collection, obtains the SVM of training completion
Classifier;
The testing feature vector collection is inputted in the SVM classifier that the training is completed and is tested, test knot is obtained
Fruit.
Preferably, it is DeepID network that the preset features, which extract network,.
Preferably, described that the facial characteristics that network extracts preset human face data collection is extracted by preset features, it obtains by face
The facial characteristics vector set that feature vector is constituted, before further include:
Multiple characteristic points on facial image are extracted, face characteristic point diagram is obtained;
The face characteristic point diagram is successively carried out handling based on multiple dimensioned and gray processing enhancing, after obtaining enhancing processing
Face characteristic point diagram;
The enhanced face characteristic point diagram is made into overturning processing, obtains preset human face data collection.
Preferably, described that the facial characteristics that network extracts preset human face data collection is extracted by preset features, it obtains by face
The facial characteristics vector set that feature vector is constituted, further includes:
Dimension-reduction treatment is carried out to the facial characteristics vector using PCA.
Preferably, the SVM classifier of the training feature vector collection training based on LP norm obtains training completion
SVM classifier, comprising:
Construct the unilateral loss function based on LP norm;
The objective function that SVM classifier is obtained according to the unilateral loss function, constructs the svm classifier based on LP norm
Device.
Preferably, described input the testing feature vector collection in the SVM classifier that training is completed is tested, and is obtained
Test result, further includes:
The correct number of samples in the test result is counted, by the ratio of the correct number of samples and test sample sum
It is worth the accuracy rate as test result.
The application second aspect provides a kind of face identification device based on svm classifier, comprising:
The characteristic extracting module is extracted the facial characteristics that network extracts preset human face data collection by preset features, is obtained
The facial characteristics vector set being made of facial characteristics vector;
The division module, for by the facial characteristics vector set be divided into training feature vector collection and test feature to
Quantity set;
The training module is trained the SVM classifier based on LP norm with the training feature vector collection, obtains
The SVM classifier that training is completed;
The test module, for by the testing feature vector collection input it is described training complete SVM classifier in into
Row test, obtains test result.
Preferably, further includes:
Enhance module and obtains face characteristic point diagram for extracting multiple characteristic points on facial image;
The face characteristic point diagram is successively carried out handling based on multiple dimensioned and gray processing enhancing, after obtaining enhancing processing
Face characteristic point diagram;
The enhanced face characteristic point diagram is made into overturning processing, obtains preset human face data collection.
Preferably, further includes:
Dimensionality reduction module carries out dimension-reduction treatment to the facial characteristics vector using PCA.
Preferably, further includes:
Statistical module by the correct number of samples and is surveyed for counting the correct number of samples in the test result
Try accuracy rate of the ratio of total sample number as test result.
As can be seen from the above technical solutions, the embodiment of the present application has the advantage that
In the application, a kind of face identification method based on svm classifier is provided, comprising: firstly, passing through feature extraction net
Network extracts the feature of preset human face data collection, obtains set of eigenvectors;Secondly, set of eigenvectors is divided into training feature vector
Collection and testing feature vector collection;Then, it with the SVM classifier of the training feature vector collection training based on LP norm, is instructed
Practice the SVM classifier completed;It is tested finally, the testing feature vector collection is inputted in the SVM classifier that training is completed,
Obtain test result.Face identification method provided by the present application based on svm classifier, using the SVM classifier based on LP norm
Classify to human face data collection, can effectively eliminate biggish residual error caused by exceptional data point in training process, thus
Classifier is improved to the robustness of abnormal data, solves face identification method neural network based to Complex Noise
Facial image when being identified, the poor technical problem of robustness.
Detailed description of the invention
Fig. 1 is a kind of flow diagram of the embodiment one of the face identification method based on svm classifier provided by the present application;
Fig. 2 is a kind of flow diagram of the embodiment two of the face identification method based on svm classifier provided by the present application;
Fig. 3 is a kind of structural schematic diagram of the embodiment of the face identification device based on svm classifier provided by the present application.
Specific embodiment
In order to make those skilled in the art more fully understand application scheme, below in conjunction in the embodiment of the present application
Attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is only this
Apply for a part of the embodiment, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art exist
Every other embodiment obtained under the premise of creative work is not made, shall fall in the protection scope of this application.
In order to make it easy to understand, referring to Fig. 1, a kind of reality of face identification method based on svm classifier provided by the present application
Apply example one, comprising:
Step 101 extracts the facial characteristics that network extracts preset human face data collection by preset features, obtains by facial characteristics
The facial characteristics vector set that vector is constituted.
Wherein, preset features extract the neural network that network is trained completion, are used as feature extraction in this application
Device uses;Facial characteristics not only includes the global characteristics that local feature further includes face.
Facial characteristics vector set is divided into training feature vector collection and testing feature vector collection by step 102.
Wherein, partition mechanism is to divide by preset proportion, specific preset proportion can according to actual data volume or
It needs to be configured.
Step 103 is trained the SVM classifier based on LP norm with training feature vector collection, obtains training completion
SVM classifier.
Testing feature vector collection is inputted in the SVM classifier that the training is completed and is tested by step 104, is surveyed
Test result.
It should be noted that the SVM classifier based on LP norm is compared to the existing SVM based on bilateral loss function points
Class device, more robustness, especially for the image for having Complex Noise.The existing noisy facial image of band in addition to be by
Caused by obtaining machine, also some blocked in illumination, is obtained under the non-ideal conditions such as expression and posture.SVM classifier
Set of eigenvectors is exactly projected to higher dimensional space by processing feature vector set, and optimum linearity point is then sought in the higher dimensional space
Class hyperplane carries out hyperplane division, classifies to realize to face characteristic.
The application classifies to human face data collection using the SVM classifier based on LP norm, can effectively eliminate instruction
Biggish residual error caused by exceptional data point solves base to improve classifier to the robustness of abnormal data during white silk
In neural network face identification method when to being identified with the facial image of Complex Noise, the poor technology of robustness
Problem.
In order to make it easy to understand, referring to figure 2., a kind of reality of the face identification method based on svm classifier provided by the present application
Example two is applied, by taking the processing of practical facial image as an example, the implementation procedure of the present embodiment is described in detail, comprising:
Step 201 makees enhancing processing to facial image, obtains preset human face data collection.
Firstly, extracting 5 characteristic points on facial image, it is respectively as follows: eyes two o'clock, a bit, mouth two o'clock obtains nose
5 feature point diagrams, feature point diagram here is not the characteristic pattern extracted after feature, but the characteristic point of facial image different zones
Figure, and be RGB color figure;5 feature point diagrams are extended into 10 feature point diagrams by the way of alignment;Secondly, by 10 spies
Sign point diagram carries out multi-scale transform three times, available 30 feature point diagrams respectively;Then, by 30 RGB color feature point diagrams
Gray processing enhancing is carried out, 60 feature point diagrams is obtained, is handled finally by overturning, 60 feature point diagrams are enhanced to 120, until
This, completes the enhancing processing operation of facial image, and every facial image is all used this step process, obtains preset human face data
Collection.
Step 202, the facial characteristics that preset human face data collection is extracted by DeepID network, obtain by facial characteristics vector structure
At facial characteristics vector set.
Wherein, DeepID network has 4 convolutional layers, wherein first 3 every layer of convolutional layer output are all connected to a maximum pond
Layer, activation primitive are ReLU function, and the 4th layer of convolutional layer is directly connected with last full articulamentum, and full articulamentum is by 160 nerves
The available output of face characteristic point diagram input network is the feature vector of 160 dimensions by member composition;Using full articulamentum as defeated
Layer out.
It should be noted that 120 feature point diagrams that preset human face data is integrated press overturning front and back as node, before overturning
Two pictures afterwards input a network, i.e., two DeepID features after each network output feature point diagram and overturning;
The full articulamentum of DeepID network is made of 160 neurons, therefore 120 different parts face characteristic point diagrams need to input 60
DeepID network obtains 2 × 60 DeepID features, so, every face finally can extract to 160 × 2 × 60, i.e., 19200
The facial characteristics vector of dimension;The facial characteristics vector set of last available human face data collection.
Step 203 carries out dimension-reduction treatment to the facial characteristics vector in facial set of eigenvectors, obtains set of eigenvectors.
It should be noted that carrying out dimension-reduction treatment to the facial characteristics vector in facial set of eigenvectors using PCA, obtain
Set of eigenvectors.
Set of eigenvectors is divided into training feature vector collection and testing feature vector collection by step 204.
It should be noted that processing complete set of eigenvectors according to the ratio of 4:1 be divided into training feature vector collection and
Testing feature vector collection.
Step 205 is trained the SVM classifier based on LP norm with training feature vector collection, obtains training completion
SVM classifier.
It should be noted that firstly, with X=[x1,...xn]∈Rd×nTraining feature vector collection is constructed single using LP norm
Side loss function obtains the unilateral loss function formula based on LP norm:
The formula of the known SVM classifier based on Lp norm:
Secondly, by the objective function of unilateral loss function common recognition and the SVM classifier based on LP norm based on LP norm
In conjunction with the SVM classifier objective function Equation of the available LP norm with the upper limit:
Wherein, It is regularization term, γ is balance parameters,It is projection matrix, WTx
+ b is linear model,It is bias term, yi∈ { 1, -1 } is human face data label,It is description xiIt is unilateral
The slack variable of loss function, ε are loss ceiling amounts.
It should be noted that no matter in unilateral loss functionHow is value no matter
Facial image classification deviation with Complex Noise has much, can pass through a given upper limit and restrict singular point to hyperplane
Distance, i.e. restricted part singular point causes hyperplane to divide inaccuracy far from hyperplane.Singular point can be reduced in this approach
To the influence power that hyperplane divides, while there is the exceptional data point of larger residual error and singular point can be removed.Unilateral loss letter
ε item in number is this upper limit, if the distance of singular point to hyperplane is greater than ε, is ε apart from value, therefore has upper
The loss function of limit more has robustness to data outliers.If the objective function of SVM classifier only includes loss function, can
Can meeting so that model over-fitting, and the addition of regularization term can solve this problem, to improve the generalization ability of classifier.
Testing feature vector collection is inputted in the SVM classifier that training is completed and is tested by step 206, obtains test knot
Fruit.
It should be noted that resulting test result is face classification results, each class testing represents the survey of a face
Examination, i.e., realize the identification of face in a manner of classification.
Correct number of samples in step 207, statistical test result, by the ratio of correct number of samples and test sample sum
It is worth the accuracy rate as test result.
In order to make it easy to understand, referring to figure 3., this application provides a kind of realities of face identification device based on svm classifier
Apply example, comprising: enhancing module 301, dimensionality reduction module 303, division module 304, training module 305, is surveyed characteristic extracting module 302
Die trial block 306, statistical module 307;
Enhance module 301, for extracting multiple characteristic points on facial image, obtains face characteristic point diagram;
Face characteristic point diagram is successively carried out to the enhancing based on multiple dimensioned and gray processing to handle, obtains enhancing treated people
Face characteristic point figure;
Enhanced face characteristic point diagram is made into overturning processing, obtains preset human face data collection.
Characteristic extracting module 302 extracts network by preset features and extracts the facial characteristics of preset human face data collection, obtain by
The facial characteristics vector set that facial characteristics vector is constituted.
Dimensionality reduction module 303 carries out dimension-reduction treatment to facial feature vector using PCA, obtains set of eigenvectors.
Division module 304, for set of eigenvectors to be divided into training feature vector collection and testing feature vector collection.
Training module 305 is trained the SVM classifier based on LP norm with training feature vector collection, is trained
The SVM classifier of completion.
Test module 306 is tested for inputting testing feature vector collection in the SVM classifier that training is completed, and is obtained
To test result.
Statistical module 307, for counting the correct number of samples in the test result, by the correct number of samples with
Accuracy rate of the ratio of test sample sum as test result.
In several embodiments provided herein, it should be understood that disclosed device and method can pass through it
Its mode is realized.For example, the apparatus embodiments described above are merely exemplary, for example, the division of the unit, only
Only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components can be tied
Another system is closed or is desirably integrated into, or some features can be ignored or not executed.Another point, it is shown or discussed
Mutual coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or logical of device or unit
Letter connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a computer readable storage medium.Based on this understanding, the technical solution of the application is substantially
The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words
It embodies, which is stored in a storage medium, including some instructions are to pass through a computer
Equipment (can be personal computer, server or the network equipment etc.) executes the complete of each embodiment the method for the application
Portion or part steps.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (full name in English: Read-Only
Memory, english abbreviation: ROM), random access memory (full name in English: Random Access Memory, english abbreviation:
RAM), the various media that can store program code such as magnetic or disk.
The above, above embodiments are only to illustrate the technical solution of the application, rather than its limitations;Although referring to before
Embodiment is stated the application is described in detail, those skilled in the art should understand that: it still can be to preceding
Technical solution documented by each embodiment is stated to modify or equivalent replacement of some of the technical features;And these
It modifies or replaces, the spirit and scope of each embodiment technical solution of the application that it does not separate the essence of the corresponding technical solution.
Claims (10)
1. a kind of face identification method based on svm classifier characterized by comprising
The facial characteristics that network extracts preset human face data collection is extracted by preset features, obtains the face being made of facial characteristics vector
Portion's set of eigenvectors;
The facial characteristics vector set is divided into training feature vector collection and testing feature vector collection;
The SVM classifier based on LP norm is trained with the training feature vector collection, obtains the svm classifier of training completion
Device;
The testing feature vector collection is inputted in the SVM classifier that the training is completed and is tested, test result is obtained.
2. the face identification method according to claim 1 based on svm classifier, which is characterized in that the preset features mention
Taking network is DeepID network.
3. the face identification method according to claim 1 based on svm classifier, which is characterized in that described by preset features
The facial characteristics that network extracts preset human face data collection is extracted, the facial characteristics vector set being made of facial characteristics vector is obtained,
Before further include:
Multiple characteristic points on facial image are extracted, face characteristic point diagram is obtained;
The face characteristic point diagram is successively carried out the enhancing based on multiple dimensioned and gray processing to handle, obtains enhancing treated people
Face characteristic point figure;
The enhanced face characteristic point diagram is made into overturning processing, obtains preset human face data collection.
4. the face identification method according to claim 1 based on svm classifier, which is characterized in that extracted by preset features
Network extracts the facial characteristics of preset human face data collection, obtains the facial characteristics vector set being made of facial characteristics vector, also wraps
It includes:
Dimension-reduction treatment is carried out to the facial characteristics vector using PCA.
5. the face identification method according to claim 1 based on svm classifier, which is characterized in that the training
Set of eigenvectors is trained the SVM classifier based on LP norm, obtains the SVM classifier of training completion, comprising:
Construct the unilateral loss function based on LP norm;
The objective function that SVM classifier is obtained according to the unilateral loss function, constructs the SVM classifier based on LP norm.
6. the face identification method according to claim 1 based on svm classifier, described defeated by the testing feature vector collection
Enter and tested in the SVM classifier that training is completed, obtains test result, further includes:
The correct number of samples in the test result is counted, the ratio of the correct number of samples and test sample sum is made
For the accuracy rate of test result.
7. a kind of face identification device based on svm classifier characterized by comprising characteristic extracting module, division module, instruction
Practice module, test module;
The characteristic extracting module is extracted the facial characteristics that network extracts preset human face data collection by preset features, is obtained by face
The facial characteristics vector set that portion's feature vector is constituted;
The division module, for the facial characteristics vector set to be divided into training feature vector collection and testing feature vector
Collection;
The training module is trained the SVM classifier based on LP norm with the training feature vector collection, is trained
The SVM classifier of completion;
The test module is surveyed for inputting the testing feature vector collection in the SVM classifier that the training is completed
Examination, obtains test result.
8. the face identification device according to claim 7 based on svm classifier, which is characterized in that further include: enhancing mould
Block;
The enhancing module obtains face characteristic point diagram for extracting multiple characteristic points on facial image;
The face characteristic point diagram is successively carried out the enhancing based on multiple dimensioned and gray processing to handle, obtains enhancing treated people
Face characteristic point figure;
The enhanced face characteristic point diagram is made into overturning processing, obtains preset human face data collection.
9. the face identification device according to claim 7 based on svm classifier, which is characterized in that further include: dimensionality reduction mould
Block;
The dimensionality reduction module carries out dimension-reduction treatment to the facial characteristics vector using PCA.
10. the face identification device according to claim 7 based on svm classifier, which is characterized in that further include: statistics mould
Block;
The statistical module by the correct number of samples and is surveyed for counting the correct number of samples in the test result
Try accuracy rate of the ratio of total sample number as test result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910778321.7A CN110490149A (en) | 2019-08-22 | 2019-08-22 | A kind of face identification method and device based on svm classifier |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910778321.7A CN110490149A (en) | 2019-08-22 | 2019-08-22 | A kind of face identification method and device based on svm classifier |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110490149A true CN110490149A (en) | 2019-11-22 |
Family
ID=68552847
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910778321.7A Pending CN110490149A (en) | 2019-08-22 | 2019-08-22 | A kind of face identification method and device based on svm classifier |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110490149A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112949522A (en) * | 2021-03-11 | 2021-06-11 | 重庆邮电大学 | Portrait data classification method based on support vector machine |
CN113408606A (en) * | 2021-06-16 | 2021-09-17 | 中国石油大学(华东) | Semi-supervised small sample image classification method based on graph collaborative training |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101419671A (en) * | 2008-11-10 | 2009-04-29 | 北方工业大学 | Face gender identification method based on fuzzy support vector machine |
CN103559294A (en) * | 2013-11-12 | 2014-02-05 | 中国石油大学(北京) | Construction method and device as well as sorting method and device for support vector machine sorter |
CN106203356A (en) * | 2016-07-12 | 2016-12-07 | 中国计量大学 | A kind of face identification method based on convolutional network feature extraction |
CN108491772A (en) * | 2018-03-09 | 2018-09-04 | 天津港(集团)有限公司 | A kind of face recognition algorithms and face identification device |
-
2019
- 2019-08-22 CN CN201910778321.7A patent/CN110490149A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101419671A (en) * | 2008-11-10 | 2009-04-29 | 北方工业大学 | Face gender identification method based on fuzzy support vector machine |
CN103559294A (en) * | 2013-11-12 | 2014-02-05 | 中国石油大学(北京) | Construction method and device as well as sorting method and device for support vector machine sorter |
CN106203356A (en) * | 2016-07-12 | 2016-12-07 | 中国计量大学 | A kind of face identification method based on convolutional network feature extraction |
CN108491772A (en) * | 2018-03-09 | 2018-09-04 | 天津港(集团)有限公司 | A kind of face recognition algorithms and face identification device |
Non-Patent Citations (2)
Title |
---|
刘建伟,等: "p 范数正则化支持向量机分类算法", 《自动化学报》 * |
张丹丹: "基于SVM及RF的CNN分类模型及其在人脸检测中的应用研究", 《中国优秀硕士学位论文全文库 信息科技辑》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112949522A (en) * | 2021-03-11 | 2021-06-11 | 重庆邮电大学 | Portrait data classification method based on support vector machine |
CN112949522B (en) * | 2021-03-11 | 2022-06-21 | 重庆邮电大学 | Portrait data classification method based on support vector machine |
CN113408606A (en) * | 2021-06-16 | 2021-09-17 | 中国石油大学(华东) | Semi-supervised small sample image classification method based on graph collaborative training |
CN113408606B (en) * | 2021-06-16 | 2022-07-22 | 中国石油大学(华东) | Semi-supervised small sample image classification method based on graph collaborative training |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Boughrara et al. | Facial expression recognition based on a mlp neural network using constructive training algorithm | |
Leibe et al. | Analyzing appearance and contour based methods for object categorization | |
CN105138993B (en) | Establish the method and device of human face recognition model | |
Cheng et al. | Exploiting effective facial patches for robust gender recognition | |
Barros et al. | A multichannel convolutional neural network for hand posture recognition | |
Bartlett et al. | Independent components of face images: A representation for face recognition | |
CN109492643A (en) | Certificate recognition methods, device, computer equipment and storage medium based on OCR | |
Liu et al. | Feature disentangling machine-a novel approach of feature selection and disentangling in facial expression analysis | |
CN110188795A (en) | Image classification method, data processing method and device | |
CN108921061A (en) | A kind of expression recognition method, device and equipment | |
Caroppo et al. | Comparison between deep learning models and traditional machine learning approaches for facial expression recognition in ageing adults | |
Siwek et al. | Autoencoder versus PCA in face recognition | |
CN110490149A (en) | A kind of face identification method and device based on svm classifier | |
Guo et al. | Smile detection in the wild based on transfer learning | |
Park et al. | Insect classification using Squeeze-and-Excitation and attention modules-a benchmark study | |
Tan et al. | Face recognition using the weighted fractal neighbor distance | |
Rani et al. | Emotion recognition based on facial components | |
Guest et al. | Levels of representation in a deep learning model of categorization | |
Jadhav et al. | Content based facial emotion recognition model using machine learning algorithm | |
Takalkar et al. | Improving micro-expression recognition accuracy using twofold feature extraction | |
Zhong et al. | Facial expression recognition using local feature selection and the extended nearest neighbor algorithm | |
CN111598126A (en) | Lightweight traditional Chinese medicinal material identification method | |
CN108229263A (en) | The recognition methods of target object and device, robot | |
Christabel et al. | Facial feature extraction based on local color and texture for face recognition using neural network | |
Espinel et al. | Face gesture recognition using deep-learning models |
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: 20191122 |
|
RJ01 | Rejection of invention patent application after publication |