CN109815864A - A kind of facial image age recognition methods based on transfer learning - Google Patents
A kind of facial image age recognition methods based on transfer learning Download PDFInfo
- Publication number
- CN109815864A CN109815864A CN201910027211.7A CN201910027211A CN109815864A CN 109815864 A CN109815864 A CN 109815864A CN 201910027211 A CN201910027211 A CN 201910027211A CN 109815864 A CN109815864 A CN 109815864A
- Authority
- CN
- China
- Prior art keywords
- component
- follows
- dcnn
- parameter
- aver
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
1) a kind of facial image age recognition methods based on transfer learning includes the following steps: using the improvement balanced preconditioning technique of picture luminance；2) it realizes that picture feature is extracted using depth convolutional neural networks DCNN, DCNN is trained using the method for transfer learning；3) softmax classifier is used, softmax classifier is that multiple scalar parameter values by DCNN output are mapped as a probability distribution array, a possibility that each probability is corresponding tag along sort, in depth convolutional neural networks model, selection solves parameter θ using Adam optimizer, it is established based on the classifier of DCNN by the training of face picture data set and Adam optimization object function acquisition parameter θ, prediction result of the corresponding tag along sort of largest component as the classifier in the probability distribution array that forecast period takes softmax classifier to export.The present invention is obviously improved the accuracy of facial image age identification.
Description
Technical field
The present invention relates to a kind of facial image age recognition methods, especially a kind of facial image year based on transfer learning
Age recognition methods.
Background technique
With the fast development of computer vision, pattern-recognition and biological identification technology, computer based people in recent years
The estimation of face age is increasingly taken seriously.It has an extensive computer vision application prospect, including safety detection, medical jurisprudence,
Human-computer interaction (HCI), E-customer's information management (ECRM) etc..In real life, monitoring camera and age identification system are utilized
System collaboration can effectively prevent vending machine and sell cigarette and illegal drug to minor.In social security, occur
The swindle illegal activities of Automatic Teller Machine usually occur in specific age groups, thus can be mentioned by introducing age information confirmation
Preceding prevention.In field of biometrics, the facial age estimates the important supplement as a kind of individual information, can be with iris, hand
The individual identities information such as print, DNA, fingerprint combines, to improve the overall performance of biological recognition system.In short, being based on computer
Face age estimation technique not only in many field extensive applications, it also has strong with other intellectual technology amalgamations
Feature.
Although currently having relevant face age Estimation Study both at home and abroad, it is limited to individual age and generates difference, line
The reasons such as complexity, data deficiency, the disturbing factor of information are managed, so that estimation accuracy rate is not high.Fundamentally, the age is estimated
Meter problem can be divided into two Main Branches: 1) confirming a range of age (such as 29~38 years old)；2) an exact year is obtained
Age (such as 18 years old).In practical applications, many age identification missions are usually only it needs to be determined that a range of age, and determine year
Age range ratio obtains the exact age and is more easier.
Summary of the invention
In order to be obviously improved the accuracy of facial image age identifying system, the present invention provides one kind to be based on transfer learning
The recognition methods of facial image age.
The technical solution adopted by the present invention to solve the technical problems is:
A kind of facial image age recognition methods based on transfer learning, the recognition methods include the following steps:
1) using the improvement balanced preconditioning technique of picture luminance；
2) it realizes that picture feature is extracted using depth convolutional neural networks DCNN, is trained using the method for transfer learning
DCNN, comprising the following steps:
Step 2.1: face picture data being divided into three parts: training set (60%), verifying collection (20%) and test set
(20%), and guarantee that the picture in training set does not appear in verifying collection and test set；
Step 2.2: the call parameters good DCNN of pre-training on ImageNet, using transfer learning training face figure
Sheet data guarantees parameter constant of the DCNN in addition to full articulamentum (FC), and therefore, transfer learning has only finely tuned the full articulamentum of DCNN
Parameter；
Step 2.3: in training process, as the precision and loss amount of training set persistently improve, but verify collection precision and
When loss amount no longer occurs significantly to change, it is believed that DCNN training finishes, while saving the parameter finely tuned in FC；
3) softmax classifier is used, softmax classifier is to be mapped as multiple scalar parameter values by DCNN output
A possibility that one probability distribution array, each probability is corresponding tag along sort, for training dataset
And siFor image data, yi∈ 1,2, and C }, wherein C is class label number, and N is image data quantity；softmax
Can data set features be extracted and are mapped asWherein xi∈ R, model are as follows:
Wherein, each parameter definition is as follows:
Probability distribution array；
X: data mapping set；
C: class label number；
Parameter θ=(θ1,θ2,···,θC) optimization object function and excellent established by cross entropy (cross-entropy)
Change algorithm (SGD, PMSprop, Adam etc.) to solve, optimization object function is as follows:
Wherein, each parameter definition is as follows:
N: image data number；
C: class label number；
δ: dirichlet function；
R (): regularization constraint item；
In depth convolutional neural networks model, selection solves parameter θ, the classifier based on DCNN using Adam optimizer
Parameter θ is obtained by the training of face picture data set and Adam optimization object function to establish, and takes softmax points in forecast period
Prediction result of the corresponding tag along sort of largest component as the classifier in the probability distribution array of class device output, the process table
It is shown as:
Wherein, L is the prediction tag along sort of the classifier.
Further, in the step 1), preprocessing process is as follows:
Step 1.1: extracting the value of three components respectively from original RGB image；
Step 1.2: calculating the mean value of three components, and be expressed as Raver, Gaver, Baver, the calculating process such as following table
Show:
Wherein, each parameter definition is as follows:
Raver: the mean value of R component；
Gaver: the mean value of G component；
Baver: the mean value of B component；
M: picture pixels number；
Ri: the value of the R component of ith pixel；
Gi: the value of the G component of ith pixel；
Bi: the value of the B component of ith pixel；
Step 1.3: calculating global gamma Qaver, the process is as follows:
Wherein, QaverFor global gamma；
Step 1.4: the gain coefficient of each component is calculated, the process is as follows:
Wherein, parameter definition is as follows:
Nr: R component gain coefficient；
Ng: G component gain coefficient；
Nb: B component gain coefficient；
Step 1.5: the new component of image RGB is reconfigured, the process is as follows:
Wherein, parameter definition is as follows:
R*: new R component；
G*: new G component；
B*: new B component；
Step 1.6: obtained new components R*, G*, B*Amendment is in range [0~255].For being greater than 255 component
Value, is set as 255；For the component value less than 0, it is set as 0；Component value in range is remained unchanged；
Step 1.7: picture is constructed according to revised new component.
It is not necessary pre- as picture pretreatment using provided picture luminance equalization method in the step 1)
Processing mode, but generalling use this method as preconditioning technique can be such that system performance is obviously improved；In step 2), using migration
Learning training DCNN, what we finely tuned is full articulamentum.When the higher precision of system requirements and performance, part can also be finely tuned
Parameter in convolutional layer.
Technical concept of the invention are as follows: the face picture in usual data set has excessive lightness or darkness situation, this correspondence
Extracting feature with DCNN can have an adverse effect.Therefore, we used a kind of common picture luminances to equalize method conduct
Picture preconditioning technique corrects the brightness of picture.Then, we apply the feature extraction that DCNN realizes face picture.But from
One DCNN of end-to-end training can be expended considerable time and effort, and the defect of data deficiencies can make DCNN be difficult to obtain it is good
Good performance.So we solve these problems using the method for transfer learning, core operation is only to finely tune DCNN in training
Parameter in full articulamentum and keep other parameters constant.After having trained model, we realize face with softmax classifier
The prediction at picture age takes the corresponding tag along sort L conduct of largest component in the probability distribution array of softmax classifier output
Final prediction result.
Beneficial effects of the present invention are mainly shown as: 1, being located in advance using a kind of common luminance proportion method as image
Reason technology, to eliminate picture excessively or excessively secretly to the adverse effect of DCNN training and prediction.2, using the method for transfer learning,
It solves the problems, such as to take a significant amount of time and energy and data deficiencies from end-to-end trained DCNN.
Detailed description of the invention
Fig. 1 is facial image age identification model schematic diagram；
Fig. 2 is transfer learning schematic diagram.
Specific embodiment
Present invention is further described in detail with reference to the accompanying drawing.
Referring to Figures 1 and 2, a kind of facial image age recognition methods based on transfer learning, uses the side of transfer learning
Method realizes the estimation of facial image age, therefore firstly the need of progress image preprocessing (such as Fig. 1).Then doing using transfer learning
Method (such as Fig. 2) trains image data, and the last probability distribution array largest component that we take softmax classifier to export is corresponding
Label is as final prediction result.Include the following steps:
It 1) is particularly significant by Preprocessing Technique improving image quality for facial image age identifying system
, it is both the premise that learning model extracts desirable features, is also directly affected finally predicting as a result, using a kind of common
Improve the preconditioning technique of picture luminance comprising the steps of:
Step 1.1: extracting the value of three components respectively from original RGB image；
Step 1.2: calculating the mean value of three components, and be expressed as Raver, Gaver, Baver, the calculating process such as following table
Show:
Wherein, each parameter definition is as follows:
Raver: the mean value of R component；
Gaver: the mean value of G component；
Baver: the mean value of B component；
M: picture pixels number；
Ri: the value of the R component of ith pixel；
Gi: the value of the G component of ith pixel；
Bi: the value of the B component of ith pixel；
Step 1.3: calculating global gamma Qaver, the process is as follows:
Wherein, QaverFor global gamma；
Step 1.4: the gain coefficient of each component is calculated, the process is as follows:
Wherein, parameter definition is as follows:
Nr: R component gain coefficient；
Ng: G component gain coefficient；
Nb: B component gain coefficient；
Step 1.5: the new component of image RGB is reconfigured, the process is as follows:
Wherein, parameter definition is as follows:
R*: new R component；
G*: new G component；
B*: new B component；
Step 1.6: obtained new components R*, G*, B*Amendment is in range [0~255].For being greater than 255 component
Value, is set as 255；For the component value less than 0, it is set as 0；Component value in range is remained unchanged；
Step 1.7: picture being constructed according to revised new component, pretreatment finishes；
2) we are extracted using depth convolutional neural networks (DCNN) Lai Shixian picture feature.But from end-to-end training one
DCNN will be expended considerable time and effort, while the insufficient limitation of image data makes network be difficult to obtain preferable property
Energy.In order to solve these problems, we train DCNN using the method for transfer learning, comprising the following steps:
Step 2.1: face picture data being divided into three parts: training set (60%), verifying collection (20%) and test set
(20%), and guarantee that the picture in training set does not appear in verifying collection and test set；
Step 2.2: the call parameters good DCNN of pre-training on ImageNet, using transfer learning training face figure
Sheet data guarantees parameter constant of the DCNN in addition to full articulamentum (FC), and therefore, transfer learning has only finely tuned the full articulamentum of DCNN
Parameter；
Step 2.3: in training process, as the precision and loss amount of training set persistently improve, but verify collection precision and
When loss amount no longer occurs significantly to change, it is believed that DCNN training finishes, while saving the parameter of FC fine tuning；
3) in more classification problems, softmax classifier is most generally used.Softmax classifier be by it is multiple by
The scalar parameter value of DCNN output is mapped as a probability distribution array, each probability is the possibility of corresponding tag along sort
Property, for training datasetAnd siFor image data, yi∈ 1,2, and C }, wherein C is class label
Number, N are image data quantity, and data set features can be extracted and be mapped as by softmaxWherein xi∈ R, mould
Type are as follows:
Wherein, each parameter definition is as follows:
Probability distribution array；
X: data mapping set；
C: class label number；
Parameter θ=(θ1,θ2,···,θC) optimization object function and excellent established by cross entropy (cross-entropy)
Change algorithm (SGD, PMSprop, Adam etc.) to solve.Optimization object function is as follows:
Wherein, each parameter definition is as follows:
N: image data number；
C: class label number；
δ: dirichlet function；
R (): regularization constraint item；
In depth convolutional neural networks model, Adam is common optimizer.It in picture classification problem compared to
The optimizers such as SGD, RMSprop have better performance, therefore we select to solve parameter θ using Adam optimizer, are based on DCNN
Classifier by the way that face picture data set is trained and Adam optimization object function obtains parameter θ and establishes, taken in forecast period
Prediction knot of the corresponding tag along sort of largest component as the classifier in the probability distribution array of softmax classifier output
Fruit, the procedural representation are as follows:
Wherein, L is the prediction tag along sort of the classifier.
Claims (2)
1. a kind of facial image age recognition methods based on transfer learning, which is characterized in that the recognition methods includes as follows
Step:
1) using the improvement balanced preconditioning technique of picture luminance；
2) it realizes that picture feature is extracted using depth convolutional neural networks DCNN, is trained using the method for transfer learning
DCNN, comprising the following steps:
Step 2.1: face picture data are divided into three parts: training set, verifying collection and test set, and guarantee the figure in training set
Piece does not appear in verifying collection and test set；
Step 2.2: the call parameters good DCNN of pre-training on ImageNet, using transfer learning training face picture number
According to guaranteeing parameter constant of the DCNN in addition to full articulamentum, therefore, transfer learning has only finely tuned the parameter of the full articulamentum of DCNN；
Step 2.3: in training process, as the precision and loss amount of training set persistently improve, but verifying the precision and loss of collection
When apparent variation no longer occurs for amount, it is believed that DCNN training finishes, while saving the parameter finely tuned in FC；
3) softmax classifier is used, softmax classifier is that multiple scalar parameter values by DCNN output are mapped as one
A possibility that probability distribution array, each probability is corresponding tag along sort, for training datasetAnd si
For image data, yi∈ { 1,2 ..., C }, wherein C is class label number, and N is image data quantity；Softmax can be by data
Collection feature extraction is simultaneously mapped asWherein xi∈ R, model are as follows:
Wherein, each parameter definition is as follows:
Probability distribution array；
X: data mapping set；
C: class label number；
Parameter θ=(θ1,θ2,…,θC) optimization object function and optimization algorithm are established by cross entropy solve, optimization object function
It is as follows:
Wherein, each parameter definition is as follows:
N: image data number；
C: class label number；
δ: dirichlet function；
R (): regularization constraint item；
In depth convolutional neural networks model, selection solves parameter θ using Adam optimizer, and the classifier based on DCNN passes through
Face picture data set is trained and Adam optimization object function obtains parameter θ to establish, and takes softmax classifier in forecast period
Prediction result of the corresponding tag along sort of largest component as the classifier, the procedural representation in the probability distribution array of output
Are as follows:
Wherein, L is the prediction tag along sort of the classifier.
2. a kind of facial image age recognition methods based on transfer learning as described in claim 1, which is characterized in that described
In step 1), preprocessing process is as follows:
Step 1.1: extracting the value of three components respectively from original RGB image；
Step 1.2: calculating the mean value of three components, and be expressed as Raver, Gaver, Baver, which indicates as follows:
Wherein, each parameter definition is as follows:
Raver: the mean value of R component；
Gaver: the mean value of G component；
Baver: the mean value of B component；
M: picture pixels number；
Ri: the value of the R component of ith pixel；
Gi: the value of the G component of ith pixel；
Bi: the value of the B component of ith pixel；
Step 1.3: calculating global gamma Qaver, the process is as follows:
Wherein, QaverFor global gamma；
Step 1.4: the gain coefficient of each component is calculated, the process is as follows:
Wherein, parameter definition is as follows:
Nr: R component gain coefficient；
Ng: G component gain coefficient；
Nb: B component gain coefficient；
Step 1.5: the new component of image RGB is reconfigured, the process is as follows:
Wherein, parameter definition is as follows:
R*: new R component；
G*: new G component；
B*: new B component；
Step 1.6: obtained new components R*, G*, B*Amendment is in range [0~255], for being greater than 255 component value, if
It is set to 255；For the component value less than 0, it is set as 0；Component value in range is remained unchanged；
Step 1.7: picture is constructed according to revised new component.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910027211.7A CN109815864B (en) | 2019-01-11 | 2019-01-11 | Facial image age identification method based on transfer learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910027211.7A CN109815864B (en) | 2019-01-11 | 2019-01-11 | Facial image age identification method based on transfer learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109815864A true CN109815864A (en) | 2019-05-28 |
| CN109815864B CN109815864B (en) | 2021-01-01 |
Family
ID=66603389
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910027211.7A Active CN109815864B (en) | 2019-01-11 | 2019-01-11 | Facial image age identification method based on transfer learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109815864B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110186375A (en) * | 2019-06-06 | 2019-08-30 | 西南交通大学 | Intelligent high-speed rail white body assemble welding feature detection device and detection method |
| CN110427846A (en) * | 2019-07-19 | 2019-11-08 | 西安工业大学 | It is a kind of using convolutional neural networks to the face identification method of uneven small sample |
| CN110427804A (en) * | 2019-06-18 | 2019-11-08 | 中山大学 | A kind of iris auth method based on secondary migration study |
| CN110503154A (en) * | 2019-08-27 | 2019-11-26 | 携程计算机技术（上海）有限公司 | Method, system, electronic equipment and the storage medium of image classification |
| CN113221920A (en) * | 2021-05-20 | 2021-08-06 | 北京百度网讯科技有限公司 | Image recognition method, device, equipment, storage medium and computer program product |
| CN114925802A (en) * | 2021-12-27 | 2022-08-19 | 天翼数字生活科技有限公司 | Integrated transfer learning method and system based on depth feature mapping |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104866829A (en) * | 2015-05-25 | 2015-08-26 | 苏州大学 | Cross-age face verify method based on characteristic learning |
| CN106355248A (en) * | 2016-08-26 | 2017-01-25 | 深圳先进技术研究院 | Deep convolution neural network training method and device |
| US20170193336A1 (en) * | 2014-12-22 | 2017-07-06 | Yahoo! Inc. | Generating preference indices for image content |
| CN107545245A (en) * | 2017-08-14 | 2018-01-05 | 中国科学院半导体研究所 | A kind of age estimation method and equipment |
| US20180289334A1 (en) * | 2017-04-05 | 2018-10-11 | doc.ai incorporated | Image-based system and method for predicting physiological parameters |
| US20180293429A1 (en) * | 2017-03-30 | 2018-10-11 | George Mason University | Age invariant face recognition using convolutional neural networks and set distances |
| CN108830326A (en) * | 2018-06-21 | 2018-11-16 | 河南工业大学 | A kind of automatic division method and device of MRI image |
| CN109002755A (en) * | 2018-06-04 | 2018-12-14 | 西北大学 | Age estimation model building method and estimation method based on facial image |
-
2019
- 2019-01-11 CN CN201910027211.7A patent/CN109815864B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170193336A1 (en) * | 2014-12-22 | 2017-07-06 | Yahoo! Inc. | Generating preference indices for image content |
| CN104866829A (en) * | 2015-05-25 | 2015-08-26 | 苏州大学 | Cross-age face verify method based on characteristic learning |
| CN106355248A (en) * | 2016-08-26 | 2017-01-25 | 深圳先进技术研究院 | Deep convolution neural network training method and device |
| US20180293429A1 (en) * | 2017-03-30 | 2018-10-11 | George Mason University | Age invariant face recognition using convolutional neural networks and set distances |
| US20180289334A1 (en) * | 2017-04-05 | 2018-10-11 | doc.ai incorporated | Image-based system and method for predicting physiological parameters |
| CN107545245A (en) * | 2017-08-14 | 2018-01-05 | 中国科学院半导体研究所 | A kind of age estimation method and equipment |
| CN109002755A (en) * | 2018-06-04 | 2018-12-14 | 西北大学 | Age estimation model building method and estimation method based on facial image |
| CN108830326A (en) * | 2018-06-21 | 2018-11-16 | 河南工业大学 | A kind of automatic division method and device of MRI image |
Non-Patent Citations (2)
| Title |
|---|
| XIN LIU ET AL: "AgeNet: Deeply Learned Regressor and Classifier for", 《2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOP》 * |
| 曹 檑 等: "基于深度卷积网络的人脸年龄分析算法与实现", 《软件工程》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110186375A (en) * | 2019-06-06 | 2019-08-30 | 西南交通大学 | Intelligent high-speed rail white body assemble welding feature detection device and detection method |
| CN110427804A (en) * | 2019-06-18 | 2019-11-08 | 中山大学 | A kind of iris auth method based on secondary migration study |
| CN110427804B (en) * | 2019-06-18 | 2022-12-09 | 中山大学 | Iris identity verification method based on secondary transfer learning |
| CN110427846A (en) * | 2019-07-19 | 2019-11-08 | 西安工业大学 | It is a kind of using convolutional neural networks to the face identification method of uneven small sample |
| CN110427846B (en) * | 2019-07-19 | 2022-12-06 | 西安工业大学 | Face recognition method for small unbalanced samples by using convolutional neural network |
| CN110503154A (en) * | 2019-08-27 | 2019-11-26 | 携程计算机技术（上海）有限公司 | Method, system, electronic equipment and the storage medium of image classification |
| CN113221920A (en) * | 2021-05-20 | 2021-08-06 | 北京百度网讯科技有限公司 | Image recognition method, device, equipment, storage medium and computer program product |
| CN113221920B (en) * | 2021-05-20 | 2024-01-12 | 北京百度网讯科技有限公司 | Image recognition method, apparatus, device, storage medium, and computer program product |
| CN114925802A (en) * | 2021-12-27 | 2022-08-19 | 天翼数字生活科技有限公司 | Integrated transfer learning method and system based on depth feature mapping |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109815864B (en) | 2021-01-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109815864A (en) | A kind of facial image age recognition methods based on transfer learning | |
| EP3872650A1 (en) | Method for footprint image retrieval | |
| CN104866829B (en) | A kind of across age face verification method based on feature learning | |
| CN103578119B (en) | Target detection method in Codebook dynamic scene based on superpixels | |
| CN109829443A (en) | Video behavior recognition methods based on image enhancement Yu 3D convolutional neural networks | |
| CN109101938B (en) | Multi-label age estimation method based on convolutional neural network | |
| CN101539990B (en) | Method for selecting and rapidly comparing robust features of iris images | |
| CN102902956B (en) | A kind of ground visible cloud image identifying processing method | |
| CN106228137A (en) | A kind of ATM abnormal human face detection based on key point location | |
| CN108491797A (en) | A kind of vehicle image precise search method based on big data | |
| CN104134068B (en) | Monitoring vehicle feature representation and classification method based on sparse coding | |
| CN110348416A (en) | Multi-task face recognition method based on multi-scale feature fusion convolutional neural network | |
| CN110929679A (en) | Non-supervision self-adaptive pedestrian re-identification method based on GAN | |
| CN109684922A (en) | A kind of recognition methods based on the multi-model of convolutional neural networks to finished product dish | |
| CN107203752A (en) | A kind of combined depth study and the face identification method of the norm constraint of feature two | |
| CN108537143B (en) | A kind of face identification method and system based on key area aspect ratio pair | |
| CN110675421B (en) | Depth image collaborative segmentation method based on few labeling frames | |
| CN110472518B (en) | Fingerprint image quality judgment method based on full convolution network | |
| CN109934203A (en) | A kind of cost-sensitive increment type face identification method based on comentropy selection | |
| CN109492528A (en) | A kind of recognition methods again of the pedestrian based on gaussian sum depth characteristic | |
| CN107038416A (en) | A kind of pedestrian detection method based on bianry image modified HOG features | |
| CN109726703A (en) | A kind of facial image age recognition methods based on improvement integrated study strategy | |
| CN108564040A (en) | A kind of fingerprint activity test method based on depth convolution feature | |
| CN107220598A (en) | Iris Texture Classification based on deep learning feature and Fisher Vector encoding models | |
| CN111126155B (en) | Pedestrian re-identification method for generating countermeasure network based on semantic constraint |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |