CN109919133A - A kind of Pose-varied face recognition method based on convolutional neural networks - Google Patents
A kind of Pose-varied face recognition method based on convolutional neural networks Download PDFInfo
- Publication number
- CN109919133A CN109919133A CN201910224956.2A CN201910224956A CN109919133A CN 109919133 A CN109919133 A CN 109919133A CN 201910224956 A CN201910224956 A CN 201910224956A CN 109919133 A CN109919133 A CN 109919133A
- Authority
- CN
- China
- Prior art keywords
- neural networks
- convolutional neural
- pose
- face recognition
- method based
- 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
Abstract
The Pose-varied face recognition method based on convolutional neural networks that the invention discloses a kind of, comprise the steps of: A, based on convolutional neural networks model, change the number of the convolution kernel of convolutional layer and the size of convolution kernel, and tested on the face database of open source, obtain its optimal pond mode；B, change the pond mode of pond layer, and tested on the face database of open source, obtain its change mode；C, parameter is handled in a manner of regularization；D, spatial pyramid pondization is applied in convolutional neural networks, the present invention is based on the Pose-varied face recognition methods of convolutional neural networks to carry out recognition of face using convolutional neural networks, not only feature extraction efficiency is high, the simplification of data format, while the accuracy rate identified is high.
Description
Technical field
The present invention relates to technical field of face recognition, specifically a kind of Pose-varied face recognition based on convolutional neural networks
Method.
Background technique
Nowadays, with the fast development of the correlation theory of computer vision and application study, computer vision technique exists
The superiority applied in daily life also increasingly highlights.Carrying out identification to image with computer is computer from relevant view
Corresponding feature is extracted in frequency or image sequence, to allow the content of computer " understanding " image, and the skill that can correctly classify
Art.The promotion of security protection consciousness also allows people constantly soaring for public and personal demand for security, so that computer vision exists
Recognition of face, Face datection etc. have very high application value.
Face recognition technology is a kind of biological identification technology for carrying out identification based on facial feature information of people, is utilized
The vision system of computer mould apery class is extracted the feature of face by computer, and carries out identity according to the feature of extraction and test
Card.As an important branch of living things feature recognition, recognition of face increasingly becomes present mode identification and leads with artificial intelligence
The research hotspot in domain.In Baidu's annual meeting in 2012, Li Yanhong specially refers to Baidu's knowledge figure；Figure searches the accuracy rate of figure from 20%
It is promoted to 80%, Li Yanhong and this be obviously improved is attributed to just online face recognition search.Major mouth such as Google, search dog
Family website also all online face recognition search.In April in the same year, with the appearance of Google glass, various fantasies are being committed to
In practice, the product of reality ＂ is expanded as a ＂, the various functions of smart phone may be implemented, if sound control is taken pictures, depending on
Frequency converses and processing text information and Email etc., wherein exciting is exactly face identification functions.Smart phone
Face unlock is also provided with one of defence line for our information security.With domestic and international major network company's future development strategy meter
The formulation drawn and smart phone it is universal, recognition of face also further incorporates daily life, closely bound up with our life.
Influence due to recognition of face effect vulnerable to many factors, and requirement of the convolutional neural networks for data is without that
It is harsh, and discrimination is higher.So convolutional neural networks have the research of field of face identification important theoretical meaning
Justice and realistic meaning.
Summary of the invention
The Pose-varied face recognition method based on convolutional neural networks that the purpose of the present invention is to provide a kind of, to solve
State the problem of proposing in background technique.
In order to achieve the object, the invention provides the following technical scheme:
A kind of Pose-varied face recognition method based on convolutional neural networks comprising the steps of:
A, it is based on convolutional neural networks model, changes the number of the convolution kernel of convolutional layer and the size of convolution kernel, and increasing income
Face database on tested, obtain its parameter；
B, change the pond mode of pond layer, and tested on the face database of open source, obtain its optimal Chi Huafang
Formula；
C, parameter is handled in a manner of regularization；
D, spatial pyramid pondization is applied in convolutional neural networks；
E, result verification.
As further technical solution of the present invention: the algorithm simulating in step F passes through MATLAB software realization.
As further technical solution of the present invention: the formed precision refers between final product and part mathematical model
Error size, including form accuracy, dimensional accuracy and the aspect of surface roughness three.
As further technical solution of the present invention: the step E is specifically: by the face database of open source, to this
The innovatory algorithm of project is trained and tests, and compares other recognizers, analyzes respective advantage and disadvantage and generates such phenomenon
The reason of, the feasibility and correctness of verification algorithm.
As further technical solution of the present invention: further including step F: carrying out algorithm simulating, and be improved identification
Algorithm is built experiment porch and is verified；
As further technical solution of the present invention: further including step G: realizing face recognition algorithms using C language, carry out practical
Using
Compared with prior art, the beneficial effects of the present invention are: the present invention is based on the Pose-varied face recognitions of convolutional neural networks
Method carries out recognition of face using convolutional neural networks, and not only feature extraction efficiency is high, and data format is simple, while the standard identified
True rate is high.
Specific embodiment
The technical scheme in the embodiments of the invention will be clearly and completely described below, it is clear that described implementation
Example is only a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, this field is common
Technical staff's every other embodiment obtained without making creative work belongs to the model that the present invention protects
It encloses.
A kind of embodiment 1: Pose-varied face recognition method based on convolutional neural networks comprising the steps of:
A, it is based on convolutional neural networks model, changes the number of the convolution kernel of convolutional layer and the size of convolution kernel, and increasing income
Face database on tested, obtain its optimal parameter.
B, change the pond mode of pond layer, and tested on the face database of open source, obtain its optimal pond
Change mode.
C, the mode for learning regularization handles parameter, avoids the occurrence of over-fitting and poor fitting phenomenon, enhances network
Generalization ability and sparsity.
D, spatial pyramid pondization is applied in convolutional neural networks, enhances the flexibility and validity of network, solved
Limitation of the convolutional neural networks for input image size.
E, result verification:
F, algorithm simulating is carried out using MATLAB software, and builds experiment porch for improved recognizer and is verified.
G, face recognition algorithms are realized using C language or other language, carries out practical application.
Embodiment 2: on the basis of embodiment 1, step E is specifically the face database by open source, to this project
The reason of innovatory algorithm is trained and tests, and compares other recognizers, analyzes respective advantage and disadvantage and generates such phenomenon,
The feasibility and correctness of verification algorithm.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (5)
1. a kind of Pose-varied face recognition method based on convolutional neural networks, which is characterized in that comprise the steps of:
Based on convolutional neural networks model, change the number of the convolution kernel of convolutional layer and the size of convolution kernel, and in open source
It is tested on face database, obtains its parameter；
Change the pond mode of pond layer, and tested on the face database of open source, obtains its optimal pond mode；
Parameter is handled in a manner of regularization；
Spatial pyramid pondization is applied in convolutional neural networks；
Result verification.
2. a kind of Pose-varied face recognition method based on convolutional neural networks according to claim 1, which is characterized in that
Algorithm simulating in step F passes through MATLAB software realization.
3. a kind of Pose-varied face recognition method based on convolutional neural networks according to claim 1, which is characterized in that
The step E is specifically: by the face database of open source, innovatory algorithm is trained and is tested, while verification algorithm
Feasibility and correctness.
4. a kind of Pose-varied face recognition method based on convolutional neural networks according to claim 1, which is characterized in that
Further include step F: carrying out algorithm simulating, and build experiment porch for improved recognizer and verified.
5. a kind of Pose-varied face recognition method based on convolutional neural networks according to claim 4, which is characterized in that
Further include step G: realizing face recognition algorithms using C language, carry out practical application.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910224956.2A CN109919133A (en) | 2019-03-24 | 2019-03-24 | A kind of Pose-varied face recognition method based on convolutional neural networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910224956.2A CN109919133A (en) | 2019-03-24 | 2019-03-24 | A kind of Pose-varied face recognition method based on convolutional neural networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109919133A true CN109919133A (en) | 2019-06-21 |
Family
ID=66966379
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910224956.2A Pending CN109919133A (en) | 2019-03-24 | 2019-03-24 | A kind of Pose-varied face recognition method based on convolutional neural networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109919133A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105868774A (en) * | 2016-03-24 | 2016-08-17 | 西安电子科技大学 | Selective search and convolutional neural network based vehicle logo recognition method |
| CN107944442A (en) * | 2017-11-09 | 2018-04-20 | 北京智芯原动科技有限公司 | Based on the object test equipment and method for improving convolutional neural networks |
| CN108108676A (en) * | 2017-12-12 | 2018-06-01 | 北京小米移动软件有限公司 | Face identification method, convolutional neural networks generation method and device |
-
2019
- 2019-03-24 CN CN201910224956.2A patent/CN109919133A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105868774A (en) * | 2016-03-24 | 2016-08-17 | 西安电子科技大学 | Selective search and convolutional neural network based vehicle logo recognition method |
| CN107944442A (en) * | 2017-11-09 | 2018-04-20 | 北京智芯原动科技有限公司 | Based on the object test equipment and method for improving convolutional neural networks |
| CN108108676A (en) * | 2017-12-12 | 2018-06-01 | 北京小米移动软件有限公司 | Face identification method, convolutional neural networks generation method and device |
Non-Patent Citations (2)
| Title |
|---|
| KAIMING HE ET AL: "Spatial pyramid pooling in deep convolutional networks for visual recognition", 《IEEE TRANSACTIONS ON PATTERN ANALYSIS & MACHINE INTELLIGENCE》 * |
| 韩东 等: "基于改进的卷积神经网络多姿态人脸识别研究", 《吉林大学学报（信息科学版）》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| William et al. | Face recognition using facenet (survey, performance test, and comparison) | |
| US10282530B2 (en) | Verifying identity based on facial dynamics | |
| CN102663370B (en) | Face identification method and system | |
| CN113610540B (en) | River crab anti-counterfeiting tracing method and system | |
| Khosravy et al. | Model inversion attack: analysis under gray-box scenario on deep learning based face recognition system | |
| Xu et al. | Research on inception module incorporated siamese convolutional neural networks to realize face recognition | |
| CN105740808B (en) | Face identification method and device | |
| CN108960342A (en) | Based on the image similarity calculation method for improving SoftMax loss function | |
| Sheng et al. | Adaptive semantic-spatio-temporal graph convolutional network for lip reading | |
| CN106056074A (en) | Single training sample face identification method based on area sparse | |
| Pandey et al. | Face Recognition Using Machine Learning | |
| CN103544468B (en) | 3D facial expression recognizing method and device | |
| Singh et al. | Efficient face identification and authentication tool for biometric attendance system | |
| Lo et al. | Facial chirality: From visual self-reflection to robust facial feature learning | |
| Yi et al. | Dual model medical invoices recognition | |
| Barnachon et al. | Human actions recognition from streamed motion capture | |
| Huang et al. | Recognition method for stone carved calligraphy characters based on a convolutional neural network | |
| Liang et al. | Controller fatigue state detection based on ES-DFNN | |
| Li et al. | Effective attention-based feature decomposition for cross-age face recognition | |
| Ni | Face recognition based on deep learning under the background of big data | |
| Prihasto et al. | A survey of deep face recognition in the wild | |
| Latha et al. | A novel method for person authentication using retinal images | |
| CN109919133A (en) | A kind of Pose-varied face recognition method based on convolutional neural networks | |
| Zhao et al. | Necessary morphological patches extraction for automatic micro-expression recognition | |
| CN110390268B (en) | Three-dimensional palmprint recognition method based on geometric characteristics and direction characteristics |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190621 |
|
| WD01 | Invention patent application deemed withdrawn after publication |