CN109815801A - Face identification method and device based on deep learning - Google Patents
Face identification method and device based on deep learning Download PDFInfo
- Publication number
- CN109815801A CN109815801A CN201811550968.6A CN201811550968A CN109815801A CN 109815801 A CN109815801 A CN 109815801A CN 201811550968 A CN201811550968 A CN 201811550968A CN 109815801 A CN109815801 A CN 109815801A
- Authority
- CN
- China
- Prior art keywords
- training
- face recognition
- deep learning
- recognition model
- sample
- 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
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000013135 deep learning Methods 0.000 title claims abstract description 52
- 238000012549 training Methods 0.000 claims abstract description 102
- 238000000605 extraction Methods 0.000 claims abstract description 29
- 230000006870 function Effects 0.000 claims abstract description 29
- 239000000523 sample Substances 0.000 claims description 43
- 239000013598 vector Substances 0.000 claims description 29
- 239000013074 reference sample Substances 0.000 claims description 24
- 238000004364 calculation method Methods 0.000 claims description 23
- 238000013527 convolutional neural network Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 2
- 230000001815 facial effect Effects 0.000 abstract description 3
- 230000004913 activation Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 210000000887 face Anatomy 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 210000001331 nose Anatomy 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
A kind of face identification method based on deep learning, including two aspect of human face recognition model training method and human face recognition model application method.The basic thought of model training is the difference value reduced between similar sample, increases the difference value between non-similar sample, is accurately identified to reach to face with this.Feature extraction network is trained first, after the completion of model training, carries out disaggregated model training, softmax classification layer, the training for sorter network are added after trained feature extraction network.Classify in softmax layers to feature, loss function after training, obtains human face recognition model to reduce the distance between output and label as target.Human face recognition model application method part, is input to human face recognition model for facial image, obtains softmax classification results, classification representated by numeric score highest dimension serial number is recognition result.
Description
Technical Field
The invention relates to the field of computer vision, in particular to a face recognition method based on deep learning.
Background
With the development of deep learning technology and the rapid improvement of computer computing capability, the fields of artificial intelligence, computer vision, image processing and the like are rapidly developed. The human face recognition is a classic subject in the field of computer vision, has great research and application values, and has high social requirements. The main problem solved by face recognition is that for inputting an image containing a face, an algorithm can automatically extract features of the face on the image, recognize the face image according to the features and give a recognition result.
Currently, face recognition related technologies have advanced greatly and have seen some successful applications. Conventional face recognition methods include methods based on geometric features, methods based on principal component analysis, methods based on wavelet transformation and pattern matching, and the like. The method based on the geometric features is to perform geometric relation operation on key points (eyes, mouths, noses and the like) of a face, such as distance, angles and the like, and recognize the face according to the distance or angle relation, so that local fine features are easily ignored, the feature loss is caused, the precision is low, and the method is more suitable for rough classification; the method based on principal component analysis is also called PCA, the basic idea is from the statistical point of view, the basic elements of the distribution of the face image, namely the characteristic vector of the face image sample set covariance matrix, are searched, so as to approximately represent the face image, and the method has great limitation on the similarity requirement of the training image and the test image; the method based on wavelet transformation and graph matching accurately extracts facial feature points and a matching algorithm based on a Gabor engine, has better accuracy, can eliminate changes caused by facial gestures, expressions, hairstyles, glasses, lighting environments and the like, and has lower recognition speed.
In recent years, with the development of computer vision and deep learning, the research of deep learning in face recognition has gained more attention. Some methods for detecting and identifying human faces by using deep learning appear, but the accuracy rate of human face identification is still not high, and the increasingly improved use requirements cannot be met.
Disclosure of Invention
Aiming at the problems, the invention provides a face recognition model training method and device based on deep learning and a face recognition method and device based on deep learning, so as to solve the problems of low face recognition accuracy and low recognition speed and expand the application range of the system.
In order to solve the technical problem, according to an aspect of the present invention, there is provided a face recognition model training method based on deep learning, the method including the following steps:
1) acquiring face image data;
2) in each iterative training, N personal face images are randomly selected as training data and divided into three parts which are respectively used as a reference sample, a positive sample and a negative sample;
3) inputting the training data into a convolutional neural network for calculation, and obtaining a characteristic value at the last layer of the convolutional neural network;
4) calculating a first Loss function value Loss1, wherein the Loss1 value is formed by reducing the distance between similar samples and increasing the distance between non-similar samples, so as to realize the training of the feature extraction network;
5) after the training of the feature extraction network is finished, adding a softmax classification layer after the trained feature extraction network for fine tuning learning; training the classification network is realized;
6) and obtaining the face recognition model based on deep learning through the training of the steps.
Preferably, for the selection of the three types of samples, the selection of the reference sample is a random selection, the positive sample is a sample with the same label as the reference sample, and the selection of the negative sample follows the following formula on the basis of the reference sample and the positive sample:
whereinRepresents the characteristic value of the reference sample,represents the characteristic value of the positive sample,represents a negative sample characteristic value, i.e. the distance between the selected negative sample and the reference sample is larger than the distance between the selected positive sample and the reference sample.
Preferably, the calculation formula of the first Loss function value Loss1 is:
wherein,represents the characteristic value of the reference sample,represents the characteristic value of the positive sample,representing a negative sample characteristic value, α representing a threshold, N representing the number of samples, if saidAnd if the Loss1 value is converged, finishing the model training, and if the Loss1 value is not converged, performing parameter adjustment and iterative training.
Preferably, during the training of the classification network, in each iterative training, N training samples are randomly selected, the training data are input to the trained feature extraction network for feature extraction, calculation is performed through a softmax layer, meanwhile, second loss function calculation is performed according to label information, reverse error propagation is performed, parameters of the classification layer are adjusted, and after the network is converged, the training is finished.
Preferably, in the classification network training, the second loss function is a cross entropy average value of the softmax layer output vector and the label information.
According to another aspect of the present invention, there is provided a face recognition method based on deep learning, the method comprising the steps of:
1) the deep learning based face recognition model obtained by training according to the deep learning based face recognition model training method of claim 1;
2) inputting the face image to be detected into the trained face recognition model based on deep learning, and obtaining a classification information output vector after calculation;
3) and for the classified information output vector, the class corresponding to the highest-dimensional ordinal of the probability value is taken, and the class is the face recognition result.
Preferably, the source of the face image to be detected is a photo file and/or a camera device.
According to another aspect of the present invention, there is provided a face recognition model training device based on deep learning, including:
1) a model image acquisition device for acquiring face image data;
2) the model sample classification device randomly selects N personal face images as training data in each iterative training, and divides the N personal face images into three parts which are respectively used as a reference sample, a positive sample and a negative sample;
3) the model characteristic value acquisition device inputs the training data into a convolutional neural network for calculation, and characteristic values are obtained at the last layer of the convolutional neural network;
4) the model feature extraction network training device is used for calculating a first Loss function value Loss1, and the Loss1 value is formed by aiming at reducing the distance between similar samples and increasing the distance between non-similar samples, so that the feature extraction network is trained;
5) the model classification network training device is used for performing fine tuning learning by adding a softmax classification layer after the trained feature extraction network is finished after the feature extraction network is trained; training the classification network is realized;
6) and the face recognition model acquisition device obtains a face recognition model based on deep learning through the training of the steps.
According to another aspect of the present invention, there is provided a face recognition apparatus based on deep learning, including:
1) the acquisition device is used for acquiring the deep learning-based face recognition model obtained by training by adopting the deep learning-based face recognition model training method as claimed in claim 1;
2) the classified information output device is used for inputting the face image to be detected into the trained face recognition model based on deep learning, and obtaining a classified information output vector after calculation;
3) and the judging device is used for taking the category corresponding to the highest-dimensional ordinal of the probability value of the classification information output vector as a face recognition result.
According to another aspect of the present invention, there is provided a face recognition apparatus based on deep learning, characterized in that,
the method comprises the following steps: a processor;
and a memory having stored therein computer program instructions that, when executed by the processor, cause the processor to perform the steps of:
1) the deep learning based face recognition model obtained by training according to the deep learning based face recognition model training method of claim 1;
2) inputting the face image to be detected into the trained face recognition model based on deep learning, and obtaining a classification information output vector after calculation;
3) and for the classified information output vector, the class corresponding to the highest-dimensional ordinal of the probability value is taken, and the class is the face recognition result.
The invention provides a face recognition method based on deep learning, which adopts a deep learning method and comprises two aspects of a face recognition model training method and a face recognition model using method. The basic idea of model training is to reduce the difference value between similar samples and increase the difference value between non-similar samples, so as to achieve rapid and accurate recognition of human faces. The training method part of the face recognition model comprises the steps of feature extraction network training and classification network training which are sequentially carried out, for labeled face data, in each iteration training, N samples are randomly selected as training data and divided into three parts which are respectively used as reference samples, positive samples and negative samples, the training data are input into a convolutional neural network to carry out forward calculation, and feature vectors of all samples are obtained at the last layer of the convolutional neural network. The loss function is determined by Euclidean distance between the positive and negative samples and the characteristic vector of the reference sample, if the Euclidean distance is small, the difference is considered to be small, if the Euclidean distance is large, the difference is considered to be large, and the network parameter adjustment aims to reduce the distance between the network parameter adjustment and the positive sample and increase the distance between the network parameter adjustment and the negative sample. And obtaining a feature extraction network model. And after the characteristic extraction network model is obtained, carrying out classification model training, and adding a softmax classification layer after the trained characteristic extraction network for training the classification network. And in each iterative training, images are selected at will to serve as training data, the training data are input into the trained model to extract characteristic values, the characteristics are classified in the softmax layer, the loss function aims at reducing the distance between the output and the label, and after the training is finished, the face recognition model is obtained. And the use method part of the face recognition model inputs the face image into the face recognition model to obtain a softmax classification result, wherein the numerical score highest dimensionality sequence number is the recognition result.
The invention has the following advantages:
1) for the traditional loss function, the emphasis is to map the same type of face images to the same space, and for the method provided by the invention, the emphasis is simultaneously placed on separating different types of face images to different feature spaces on the basis, so that the improvement of the face recognition precision is greatly facilitated compared with the traditional method.
2) Compared with the traditional method, the training is simple, and the recognition speed is higher.
Drawings
The above and other objects, features and advantages of the present invention will become more apparent from the detailed description of the embodiments of the present invention when taken in conjunction with the accompanying drawings.
FIG. 1 is a flow chart of a face recognition model training method based on deep learning;
FIG. 2 is a diagram of a convolutional neural network architecture;
FIG. 3 is a flow chart of a method for using a face recognition model based on deep learning.
Detailed Description
Embodiments of the present invention are described below with reference to the drawings. It is to be understood that the specific embodiments described herein are for purposes of illustration only and are not to be construed as limitations of the invention. It should be noted that, for convenience of description, only the portions related to the present invention are shown in the drawings. Meanwhile, it should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
As shown in fig. 1, the method for training a face recognition model based on deep learning of the present invention includes the following steps:
step S11, training the feature extraction network, randomly selecting N training images in the training data set as the training batch, inputting the training batch into a convolutional neural network for convolution calculation, where the convolutional neural network has a structure as shown in fig. 2, and obtaining feature vectors at the last full connection layer.
In the substep S111, in the step S11, for the selection of N training images, special attention needs to be paid to the selection of the same type of sample, that is, for any one of the training images in the batch, it can be ensured that at least another same type of sample exists in the batch as much as possible.
And step S12, screening reference samples in the N calculated characteristic vectors, regarding any sample in the batch, if the same type of sample exists in the batch, taking the sample as the reference sample, selecting N characteristic vectors as the reference sample, taking the sample of the same type as the reference sample, and calculating the residual characteristic vectors by the following formula to obtain corresponding negative samples, thereby forming N groups of characteristic vectors, wherein each group comprises one reference sample, one positive sample and one negative sample.
Wherein,represents the characteristic value of the reference sample,represents the characteristic value of the positive sample,representing the negative sample characteristic value and N representing the number of samples.
And step S13, calculating a first loss function value for each group of vectors, performing reverse error propagation, and adjusting network parameters by using a random gradient descent algorithm. The first loss function is constructed as follows:
where each symbol has the meaning as above, α represents the threshold.
And judging whether the neural network converges or not according to the first loss function value obtained by calculation, if so, ending the feature extraction network training, and if not, returning to S11 for circular execution.
And step S14, after the training of the feature extraction network is finished, carrying out classification network training, adding a softmax classification layer after the trained feature extraction network, fixing the trained network parameters, and only adjusting the last layer of parameters in the classification network training.
And step S15, randomly selecting N samples, inputting the N samples into the trained feature extraction network, and obtaining N feature vectors.
And step S16, inputting the obtained feature vector into a classification layer, obtaining a classification result vector after calculation, firstly using a softmax activation function to perform one-hot coding processing, calculating a second loss function value according to the result, performing reverse error propagation, and adjusting classification layer network parameter values. The second loss function is as follows:
where Loss2 is a second Loss function,in order to be a prediction value of the j-th class,the j-th real value and n is the number of categories.
In the sub-step S161, regarding the step S16, the one-hot encoding process is to perform softmax activation function calculation on the network output before performing the loss function calculation, where the activation function formula is:
wherein S isiRepresents the value, V, of the current dimension calculated by the softmax activation functioniRepresenting the output value of the current dimension,representing the sum of all dimension values raised to the power of the exponent. Namely, the correct probability of prediction output of each dimension is obtained after one-hot coding processing.
Step S17, judging whether the network is convergent according to the loss function value, if so, ending the training to obtain a face recognition model; if not, the process returns to S15 for loop execution.
The design of the first Loss function and the second Loss function is very important, different types of face images can be separated into different feature spaces by adopting the first Loss function Loss1 and the second Loss function Loss2, so that the training of a model can be completed quickly and accurately, the guarantee is provided for improving the accuracy of face recognition, the training is simple, and the recognition speed is high. The application range of the face detection and recognition system is greatly widened.
As shown in fig. 2, the method for using a face recognition model based on deep learning of the present invention includes the following steps:
and S21, inputting the face image to be detected into the trained face recognition model, and obtaining an output vector of classification information after calculation.
And S22, for the output vector, the category corresponding to the highest-dimensional ordinal of the probability value is taken as the face recognition result.
For example, a vector shape is obtained at the classification level as:
[0.0012 0.03620.0267 0.0664 0.7724 0.0971]
and taking the category corresponding to the 0.7724 dimensionality 5 dimension with the maximum probability as the current face image recognition result.
So far, the technical solutions of the present invention have been described with reference to the preferred embodiments shown in the drawings, but it should be understood by those skilled in the art that the above embodiments are only for clearly illustrating the present invention, and not for limiting the scope of the present invention, and it is apparent that the scope of the present invention is not limited to these specific embodiments. Equivalent changes or substitutions of related technical features can be made by those skilled in the art without departing from the principle of the invention, and the technical scheme after the changes or substitutions can fall into the protection scope of the invention.
Claims (10)
1. A face recognition model training method based on deep learning is characterized by comprising the following steps:
1) acquiring face image data;
2) in each iterative training, N personal face images are randomly selected as training data and divided into three parts which are respectively used as a reference sample, a positive sample and a negative sample;
3) inputting the training data into a convolutional neural network for calculation, and obtaining a characteristic value at the last layer of the convolutional neural network;
4) calculating a first Loss function value Loss1, wherein the Loss1 value is formed by reducing the distance between similar samples and increasing the distance between non-similar samples, so as to realize the training of the feature extraction network;
5) after the training of the feature extraction network is finished, adding a softmax classification layer after the trained feature extraction network for fine tuning learning; training the classification network is realized;
6) and obtaining the face recognition model based on deep learning through the training of the steps.
2. The deep learning-based face recognition model training method according to claim 1, wherein for the selection of the three types of samples, the selection of the reference sample is a random selection, the positive sample is a sample with the same label as the reference sample, and the selection of the negative sample follows the following formula on the basis of the reference sample and the positive sample:
whereinRepresents the characteristic value of the reference sample,represents the characteristic value of the positive sample,represents a negative sample characteristic value, i.e. the distance between the selected negative sample and the reference sample is larger than the distance between the selected positive sample and the reference sample.
3. The deep learning based face recognition model training method according to claim 1,
the calculation formula of the first Loss function value Loss1 is:
wherein,represents the characteristic value of the reference sample,represents the characteristic value of the positive sample,representing the characteristic value of the negative sample, α representing the threshold value, and N representing the number of samples, completing model training if the Loss1 value is converged, and adjusting parameters and performing iterative training if the Loss1 value is not converged.
4. The deep learning based face recognition model training method according to claim 1,
during the classification network training, in each iterative training, N training samples are randomly selected, the training data are input into the trained feature extraction network to be subjected to feature extraction, calculation is carried out through a softmax layer, meanwhile, second loss function calculation is carried out according to label information, reverse error propagation is carried out, classification layer parameters are adjusted, and after the network is converged, the training is finished.
5. The deep learning-based face recognition model training method according to claim 4, wherein in the classification network training, the second loss function is a cross-entropy average value of the softmax layer output vector and label information.
6. A face recognition method based on deep learning is characterized by comprising the following steps:
1) the deep learning based face recognition model obtained by training according to the deep learning based face recognition model training method of claim 1;
2) inputting the face image to be detected into the trained face recognition model based on deep learning, and obtaining a classification information output vector after calculation;
3) and for the classified information output vector, the class corresponding to the highest-dimensional ordinal of the probability value is taken, and the class is the face recognition result.
7. The deep learning-based face recognition method according to claim 6, wherein the source of the face image to be detected is a photo file and/or a camera device.
8. The utility model provides a face identification model trainer based on degree of depth learning which characterized in that includes:
1) a model image acquisition device for acquiring face image data;
2) the model sample classification device randomly selects N personal face images as training data in each iterative training, and divides the N personal face images into three parts which are respectively used as a reference sample, a positive sample and a negative sample;
3) the model characteristic value acquisition device inputs the training data into a convolutional neural network for calculation, and characteristic values are obtained at the last layer of the convolutional neural network;
4) the model feature extraction network training device is used for calculating a first Loss function value Loss1, and the Loss1 value is formed by aiming at reducing the distance between similar samples and increasing the distance between non-similar samples, so that the feature extraction network is trained;
5) the model classification network training device is used for performing fine tuning learning by adding a softmax classification layer after the trained feature extraction network is finished after the feature extraction network is trained; training the classification network is realized;
6) and the face recognition model acquisition device obtains a face recognition model based on deep learning through the training of the steps.
9. A face recognition device based on deep learning is characterized by comprising:
1) the acquisition device is used for acquiring the deep learning-based face recognition model obtained by training by adopting the deep learning-based face recognition model training method as claimed in claim 1;
2) the classified information output device is used for inputting the face image to be detected into the trained face recognition model based on deep learning, and obtaining a classified information output vector after calculation;
3) and the judging device is used for taking the category corresponding to the highest-dimensional ordinal of the probability value of the classification information output vector as a face recognition result.
10. A face recognition device based on deep learning is characterized in that,
the method comprises the following steps: a processor;
and a memory having stored therein computer program instructions that, when executed by the processor, cause the processor to perform the steps of:
1) the deep learning based face recognition model obtained by training according to the deep learning based face recognition model training method of claim 1;
2) inputting the face image to be detected into the trained face recognition model based on deep learning, and obtaining a classification information output vector after calculation;
3) and for the classified information output vector, the class corresponding to the highest-dimensional ordinal of the probability value is taken, and the class is the face recognition result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811550968.6A CN109815801A (en) | 2018-12-18 | 2018-12-18 | Face identification method and device based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811550968.6A CN109815801A (en) | 2018-12-18 | 2018-12-18 | Face identification method and device based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109815801A true CN109815801A (en) | 2019-05-28 |
Family
ID=66602107
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811550968.6A Pending CN109815801A (en) | 2018-12-18 | 2018-12-18 | Face identification method and device based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109815801A (en) |
Cited By (53)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110222623A (en) * | 2019-05-31 | 2019-09-10 | 深圳市恩钛控股有限公司 | Micro- expression analysis method and system |
| CN110414550A (en) * | 2019-06-14 | 2019-11-05 | 北京迈格威科技有限公司 | Training method, device, system and the computer-readable medium of human face recognition model |
| CN110443162A (en) * | 2019-07-19 | 2019-11-12 | 南京邮电大学 | A kind of two-part training method for disguised face identification |
| CN110689066A (en) * | 2019-09-24 | 2020-01-14 | 成都考拉悠然科技有限公司 | Training method combining face recognition data equalization and enhancement |
| CN110765943A (en) * | 2019-10-23 | 2020-02-07 | 深圳市商汤科技有限公司 | Network training and recognition method and device, electronic equipment and storage medium |
| CN110781856A (en) * | 2019-11-04 | 2020-02-11 | 浙江大华技术股份有限公司 | Heterogeneous face recognition model training method, face recognition method and related device |
| CN110889489A (en) * | 2019-11-08 | 2020-03-17 | 北京小米移动软件有限公司 | Neural network training method, image recognition method and device |
| CN111079717A (en) * | 2020-01-09 | 2020-04-28 | 西安理工大学 | A face recognition method based on reinforcement learning |
| CN111126346A (en) * | 2020-01-06 | 2020-05-08 | 腾讯科技（深圳）有限公司 | Face recognition method, training method and device of classification model and storage medium |
| CN111160415A (en) * | 2019-12-12 | 2020-05-15 | 江苏艾佳家居用品有限公司 | Generative model training method for layer-by-layer inversion recognition model |
| CN111222411A (en) * | 2019-11-28 | 2020-06-02 | 中国船舶重工集团公司第七一三研究所 | Laser emission safe and rapid alarm method and device |
| CN111242199A (en) * | 2020-01-07 | 2020-06-05 | 中国科学院苏州纳米技术与纳米仿生研究所 | Image classification model training method and classification method |
| CN111259738A (en) * | 2020-01-08 | 2020-06-09 | 科大讯飞股份有限公司 | Face recognition model construction method, face recognition method and related device |
| CN111325322A (en) * | 2020-02-12 | 2020-06-23 | 深圳壹账通智能科技有限公司 | Deep learning method, system, server and storage medium based on privacy protection |
| CN111339988A (en) * | 2020-03-11 | 2020-06-26 | 福州大学 | Video face recognition method based on dynamic interval loss function and probability characteristic |
| CN111460962A (en) * | 2020-03-27 | 2020-07-28 | 武汉大学 | Mask face recognition method and system |
| CN111507188A (en) * | 2020-03-16 | 2020-08-07 | 平安科技（深圳）有限公司 | Face recognition model training method and device, computer equipment and storage medium |
| CN111582059A (en) * | 2020-04-20 | 2020-08-25 | 哈尔滨工程大学 | A face expression recognition method based on variational autoencoder |
| CN111639535A (en) * | 2020-04-29 | 2020-09-08 | 深圳英飞拓智能技术有限公司 | Face recognition method and device based on deep learning |
| CN111753641A (en) * | 2020-05-07 | 2020-10-09 | 中山大学 | A gender prediction method based on high-dimensional facial features |
| CN111753877A (en) * | 2020-05-19 | 2020-10-09 | 海克斯康制造智能技术(青岛)有限公司 | Product quality detection method based on deep neural network transfer learning |
| CN111898547A (en) * | 2020-07-31 | 2020-11-06 | 平安科技（深圳）有限公司 | Training method, device and equipment of face recognition model and storage medium |
| CN111931592A (en) * | 2020-07-16 | 2020-11-13 | 苏州科达科技股份有限公司 | Object recognition method, device and storage medium |
| CN112016501A (en) * | 2020-09-04 | 2020-12-01 | 平安科技（深圳）有限公司 | Training method and device of face recognition model and computer equipment |
| CN112052441A (en) * | 2020-08-24 | 2020-12-08 | 深圳市芯汇群微电子技术有限公司 | Data decryption method of solid state disk based on face recognition and electronic equipment |
| CN112257738A (en) * | 2020-07-31 | 2021-01-22 | 北京京东尚科信息技术有限公司 | Training method and device of machine learning model and classification method and device of image |
| CN112348045A (en) * | 2019-08-09 | 2021-02-09 | 北京地平线机器人技术研发有限公司 | Training method and training device for neural network and electronic equipment |
| CN112561076A (en) * | 2020-12-10 | 2021-03-26 | 支付宝(杭州)信息技术有限公司 | Model processing method and device |
| CN112560635A (en) * | 2020-12-10 | 2021-03-26 | 深圳云天励飞技术股份有限公司 | Face matching acceleration method and device, electronic equipment and storage medium |
| CN112733807A (en) * | 2021-02-22 | 2021-04-30 | 佳都新太科技股份有限公司 | Face comparison graph convolution neural network training method and device |
| CN112884728A (en) * | 2021-02-04 | 2021-06-01 | 复旦大学附属中山医院 | Endoscope image lesion identification screening method based on deep learning convolutional neural network |
| CN113177961A (en) * | 2021-06-07 | 2021-07-27 | 傲雄在线(重庆)科技有限公司 | Multi-mode depth model training method for seal image-text comparison |
| CN113255838A (en) * | 2021-06-29 | 2021-08-13 | 成都数之联科技有限公司 | Image classification model training method, system and device, medium and classification method |
| CN113283514A (en) * | 2021-05-31 | 2021-08-20 | 高新兴科技集团股份有限公司 | Unknown class classification method, device and medium based on deep learning |
| CN113538075A (en) * | 2020-04-14 | 2021-10-22 | 阿里巴巴集团控股有限公司 | Data processing method, model training method, device and equipment |
| CN113642541A (en) * | 2021-10-14 | 2021-11-12 | 环球数科集团有限公司 | A face attribute recognition system based on deep learning |
| CN113688762A (en) * | 2021-08-31 | 2021-11-23 | 中国平安人寿保险股份有限公司 | Face recognition method, device, equipment and medium based on deep learning |
| CN113762019A (en) * | 2021-01-22 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Feature extraction network training method, face recognition method and device |
| CN113850243A (en) * | 2021-11-29 | 2021-12-28 | 北京的卢深视科技有限公司 | Model training, face recognition method, electronic device and storage medium |
| CN113920373A (en) * | 2021-10-29 | 2022-01-11 | 平安银行股份有限公司 | Object classification method and device, terminal equipment and storage medium |
| CN113947140A (en) * | 2021-10-13 | 2022-01-18 | 北京百度网讯科技有限公司 | Training method of face feature extraction model and face feature extraction method |
| CN113963237A (en) * | 2021-12-22 | 2022-01-21 | 北京的卢深视科技有限公司 | Model training, mask-wearing state detection method, electronic equipment and storage medium |
| CN113989690A (en) * | 2021-10-28 | 2022-01-28 | 广东电网有限责任公司 | Cable classification model training method, device, equipment and storage medium |
| CN114005166A (en) * | 2021-11-25 | 2022-02-01 | 浩云科技股份有限公司 | A face recognition method and device based on multi-task learning |
| CN114065829A (en) * | 2020-08-10 | 2022-02-18 | 阿里巴巴集团控股有限公司 | Image recognition method and device and model training method and device |
| CN114241264A (en) * | 2021-12-17 | 2022-03-25 | 深圳尚米网络技术有限公司 | User discrimination model training method, user discrimination method and related device |
| CN114898454A (en) * | 2022-05-31 | 2022-08-12 | 中国第一汽车股份有限公司 | Neural network model training method and device, electronic equipment and medium |
| CN114913583A (en) * | 2022-05-17 | 2022-08-16 | 中南大学 | Image type detection method and system based on deep learning twin network |
| CN116503930A (en) * | 2023-05-11 | 2023-07-28 | 上海科江电子信息技术有限公司 | A 3D Face Recognition System with Double Sampling Mechanism and Multi-branch Parallel Training |
| CN116758374A (en) * | 2023-06-27 | 2023-09-15 | 支付宝(杭州)信息技术有限公司 | Image recognition model training method, image recognition method and device |
| CN117058440A (en) * | 2023-07-12 | 2023-11-14 | 济南博观智能科技有限公司 | A safety strap identification method, device, equipment and storage medium |
| CN117557888A (en) * | 2024-01-12 | 2024-02-13 | 清华大学深圳国际研究生院 | Face model training method and face recognition method based on metric learning |
| CN118366206A (en) * | 2024-06-19 | 2024-07-19 | 苏州元脑智能科技有限公司 | Face recognition method, face recognition device and face recognition access control system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105975959A (en) * | 2016-06-14 | 2016-09-28 | 广州视源电子科技股份有限公司 | Face feature extraction modeling and face recognition method and device based on neural network |
| CN106328121A (en) * | 2016-08-30 | 2017-01-11 | 南京理工大学 | Chinese traditional musical instrument classification method based on depth confidence network |
| CN107273876A (en) * | 2017-07-18 | 2017-10-20 | 山东大学 | A kind of micro- expression automatic identifying method of ' the grand micro- transformation models of to ' based on deep learning |
| CN108009528A (en) * | 2017-12-26 | 2018-05-08 | 广州广电运通金融电子股份有限公司 | Face authentication method, device, computer equipment and storage medium based on Triplet Loss |
| CN108875540A (en) * | 2018-03-12 | 2018-11-23 | 北京旷视科技有限公司 | Image processing method, device and system and storage medium |
| CN108875521A (en) * | 2017-12-20 | 2018-11-23 | 北京旷视科技有限公司 | Method for detecting human face, device, system and storage medium |
| CN108985377A (en) * | 2018-07-18 | 2018-12-11 | 太原理工大学 | A kind of image high-level semantics recognition methods of the multiple features fusion based on deep layer network |
-
2018
- 2018-12-18 CN CN201811550968.6A patent/CN109815801A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105975959A (en) * | 2016-06-14 | 2016-09-28 | 广州视源电子科技股份有限公司 | Face feature extraction modeling and face recognition method and device based on neural network |
| CN106328121A (en) * | 2016-08-30 | 2017-01-11 | 南京理工大学 | Chinese traditional musical instrument classification method based on depth confidence network |
| CN107273876A (en) * | 2017-07-18 | 2017-10-20 | 山东大学 | A kind of micro- expression automatic identifying method of ' the grand micro- transformation models of to ' based on deep learning |
| CN108875521A (en) * | 2017-12-20 | 2018-11-23 | 北京旷视科技有限公司 | Method for detecting human face, device, system and storage medium |
| CN108009528A (en) * | 2017-12-26 | 2018-05-08 | 广州广电运通金融电子股份有限公司 | Face authentication method, device, computer equipment and storage medium based on Triplet Loss |
| CN108875540A (en) * | 2018-03-12 | 2018-11-23 | 北京旷视科技有限公司 | Image processing method, device and system and storage medium |
| CN108985377A (en) * | 2018-07-18 | 2018-12-11 | 太原理工大学 | A kind of image high-level semantics recognition methods of the multiple features fusion based on deep layer network |
Non-Patent Citations (2)
| Title |
|---|
| THOMAS DA SILVA PAULA: "Contributions in Face Detection with Deep Neural Networks", 《PONTIFICAL CATHOLIC UNIVERSITY OF RIO GRANDE DO SUL》 * |
| 梅真卿: "复杂条件下人脸检测方法研究与实现", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (79)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110222623A (en) * | 2019-05-31 | 2019-09-10 | 深圳市恩钛控股有限公司 | Micro- expression analysis method and system |
| CN110414550B (en) * | 2019-06-14 | 2022-07-29 | 北京迈格威科技有限公司 | Training method, device and system of face recognition model and computer readable medium |
| CN110414550A (en) * | 2019-06-14 | 2019-11-05 | 北京迈格威科技有限公司 | Training method, device, system and the computer-readable medium of human face recognition model |
| CN110443162A (en) * | 2019-07-19 | 2019-11-12 | 南京邮电大学 | A kind of two-part training method for disguised face identification |
| CN110443162B (en) * | 2019-07-19 | 2022-08-30 | 南京邮电大学 | Two-stage training method for disguised face recognition |
| CN112348045A (en) * | 2019-08-09 | 2021-02-09 | 北京地平线机器人技术研发有限公司 | Training method and training device for neural network and electronic equipment |
| CN110689066A (en) * | 2019-09-24 | 2020-01-14 | 成都考拉悠然科技有限公司 | Training method combining face recognition data equalization and enhancement |
| CN110765943A (en) * | 2019-10-23 | 2020-02-07 | 深圳市商汤科技有限公司 | Network training and recognition method and device, electronic equipment and storage medium |
| CN110781856A (en) * | 2019-11-04 | 2020-02-11 | 浙江大华技术股份有限公司 | Heterogeneous face recognition model training method, face recognition method and related device |
| CN110781856B (en) * | 2019-11-04 | 2023-12-19 | 浙江大华技术股份有限公司 | Heterogeneous face recognition model training method, face recognition method and related device |
| CN110889489A (en) * | 2019-11-08 | 2020-03-17 | 北京小米移动软件有限公司 | Neural network training method, image recognition method and device |
| CN111222411A (en) * | 2019-11-28 | 2020-06-02 | 中国船舶重工集团公司第七一三研究所 | Laser emission safe and rapid alarm method and device |
| CN111222411B (en) * | 2019-11-28 | 2023-09-01 | 中国船舶重工集团公司第七一三研究所 | Laser emission safety rapid alarm method and device |
| CN111160415A (en) * | 2019-12-12 | 2020-05-15 | 江苏艾佳家居用品有限公司 | Generative model training method for layer-by-layer inversion recognition model |
| CN111160415B (en) * | 2019-12-12 | 2022-08-12 | 江苏艾佳家居用品有限公司 | Generative model training method for layer-by-layer inversion recognition model |
| CN111126346A (en) * | 2020-01-06 | 2020-05-08 | 腾讯科技（深圳）有限公司 | Face recognition method, training method and device of classification model and storage medium |
| CN111242199A (en) * | 2020-01-07 | 2020-06-05 | 中国科学院苏州纳米技术与纳米仿生研究所 | Image classification model training method and classification method |
| CN111259738A (en) * | 2020-01-08 | 2020-06-09 | 科大讯飞股份有限公司 | Face recognition model construction method, face recognition method and related device |
| CN111259738B (en) * | 2020-01-08 | 2023-10-27 | 科大讯飞股份有限公司 | Face recognition model construction method, face recognition method and related device |
| CN111079717A (en) * | 2020-01-09 | 2020-04-28 | 西安理工大学 | A face recognition method based on reinforcement learning |
| CN111325322A (en) * | 2020-02-12 | 2020-06-23 | 深圳壹账通智能科技有限公司 | Deep learning method, system, server and storage medium based on privacy protection |
| CN111339988A (en) * | 2020-03-11 | 2020-06-26 | 福州大学 | Video face recognition method based on dynamic interval loss function and probability characteristic |
| CN111339988B (en) * | 2020-03-11 | 2023-04-07 | 福州大学 | Video face recognition method based on dynamic interval loss function and probability characteristic |
| CN111507188A (en) * | 2020-03-16 | 2020-08-07 | 平安科技（深圳）有限公司 | Face recognition model training method and device, computer equipment and storage medium |
| CN111460962B (en) * | 2020-03-27 | 2023-05-05 | 武汉大学 | A mask face recognition method and system |
| CN111460962A (en) * | 2020-03-27 | 2020-07-28 | 武汉大学 | Mask face recognition method and system |
| CN113538075A (en) * | 2020-04-14 | 2021-10-22 | 阿里巴巴集团控股有限公司 | Data processing method, model training method, device and equipment |
| CN111582059A (en) * | 2020-04-20 | 2020-08-25 | 哈尔滨工程大学 | A face expression recognition method based on variational autoencoder |
| CN111582059B (en) * | 2020-04-20 | 2022-07-15 | 哈尔滨工程大学 | A face expression recognition method based on variational autoencoder |
| CN111639535B (en) * | 2020-04-29 | 2023-08-22 | 深圳英飞拓智能技术有限公司 | Face recognition method and device based on deep learning |
| CN111639535A (en) * | 2020-04-29 | 2020-09-08 | 深圳英飞拓智能技术有限公司 | Face recognition method and device based on deep learning |
| WO2021218060A1 (en) * | 2020-04-29 | 2021-11-04 | 深圳英飞拓智能技术有限公司 | Face recognition method and device based on deep learning |
| CN111753641B (en) * | 2020-05-07 | 2023-07-18 | 中山大学 | A Gender Prediction Method Based on High Dimensional Features of Face |
| CN111753641A (en) * | 2020-05-07 | 2020-10-09 | 中山大学 | A gender prediction method based on high-dimensional facial features |
| CN111753877B (en) * | 2020-05-19 | 2024-03-05 | 海克斯康制造智能技术(青岛)有限公司 | Product quality detection method based on deep neural network migration learning |
| CN111753877A (en) * | 2020-05-19 | 2020-10-09 | 海克斯康制造智能技术(青岛)有限公司 | Product quality detection method based on deep neural network transfer learning |
| CN111931592A (en) * | 2020-07-16 | 2020-11-13 | 苏州科达科技股份有限公司 | Object recognition method, device and storage medium |
| CN111898547B (en) * | 2020-07-31 | 2024-04-16 | 平安科技（深圳）有限公司 | Training method, device, equipment and storage medium for face recognition model |
| WO2021139309A1 (en) * | 2020-07-31 | 2021-07-15 | 平安科技（深圳）有限公司 | Method, apparatus and device for training facial recognition model, and storage medium |
| CN111898547A (en) * | 2020-07-31 | 2020-11-06 | 平安科技（深圳）有限公司 | Training method, device and equipment of face recognition model and storage medium |
| CN112257738A (en) * | 2020-07-31 | 2021-01-22 | 北京京东尚科信息技术有限公司 | Training method and device of machine learning model and classification method and device of image |
| CN114065829A (en) * | 2020-08-10 | 2022-02-18 | 阿里巴巴集团控股有限公司 | Image recognition method and device and model training method and device |
| CN112052441A (en) * | 2020-08-24 | 2020-12-08 | 深圳市芯汇群微电子技术有限公司 | Data decryption method of solid state disk based on face recognition and electronic equipment |
| CN112052441B (en) * | 2020-08-24 | 2021-09-28 | 深圳市芯汇群微电子技术有限公司 | Data decryption method of solid state disk based on face recognition and electronic equipment |
| CN112016501A (en) * | 2020-09-04 | 2020-12-01 | 平安科技（深圳）有限公司 | Training method and device of face recognition model and computer equipment |
| CN112016501B (en) * | 2020-09-04 | 2023-08-29 | 平安科技（深圳）有限公司 | Training method and device of face recognition model and computer equipment |
| CN112560635A (en) * | 2020-12-10 | 2021-03-26 | 深圳云天励飞技术股份有限公司 | Face matching acceleration method and device, electronic equipment and storage medium |
| CN112561076A (en) * | 2020-12-10 | 2021-03-26 | 支付宝(杭州)信息技术有限公司 | Model processing method and device |
| CN112560635B (en) * | 2020-12-10 | 2024-03-26 | 深圳云天励飞技术股份有限公司 | Face matching acceleration method and device, electronic equipment and storage medium |
| CN113762019B (en) * | 2021-01-22 | 2024-04-09 | 北京沃东天骏信息技术有限公司 | Feature extraction network training method, face recognition method and device |
| CN113762019A (en) * | 2021-01-22 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Feature extraction network training method, face recognition method and device |
| CN112884728A (en) * | 2021-02-04 | 2021-06-01 | 复旦大学附属中山医院 | Endoscope image lesion identification screening method based on deep learning convolutional neural network |
| CN112733807A (en) * | 2021-02-22 | 2021-04-30 | 佳都新太科技股份有限公司 | Face comparison graph convolution neural network training method and device |
| CN113283514A (en) * | 2021-05-31 | 2021-08-20 | 高新兴科技集团股份有限公司 | Unknown class classification method, device and medium based on deep learning |
| CN113283514B (en) * | 2021-05-31 | 2024-05-21 | 高新兴科技集团股份有限公司 | A method, device and medium for classifying unknown categories based on deep learning |
| CN113177961A (en) * | 2021-06-07 | 2021-07-27 | 傲雄在线(重庆)科技有限公司 | Multi-mode depth model training method for seal image-text comparison |
| CN113255838A (en) * | 2021-06-29 | 2021-08-13 | 成都数之联科技有限公司 | Image classification model training method, system and device, medium and classification method |
| CN113688762A (en) * | 2021-08-31 | 2021-11-23 | 中国平安人寿保险股份有限公司 | Face recognition method, device, equipment and medium based on deep learning |
| CN113688762B (en) * | 2021-08-31 | 2023-08-08 | 中国平安人寿保险股份有限公司 | Face recognition method, device, equipment and medium based on deep learning |
| CN113947140A (en) * | 2021-10-13 | 2022-01-18 | 北京百度网讯科技有限公司 | Training method of face feature extraction model and face feature extraction method |
| WO2023060936A1 (en) * | 2021-10-14 | 2023-04-20 | 环球数科集团有限公司 | Deep learning-based face attribute recognition system |
| CN113642541A (en) * | 2021-10-14 | 2021-11-12 | 环球数科集团有限公司 | A face attribute recognition system based on deep learning |
| CN113989690A (en) * | 2021-10-28 | 2022-01-28 | 广东电网有限责任公司 | Cable classification model training method, device, equipment and storage medium |
| CN113920373A (en) * | 2021-10-29 | 2022-01-11 | 平安银行股份有限公司 | Object classification method and device, terminal equipment and storage medium |
| CN114005166A (en) * | 2021-11-25 | 2022-02-01 | 浩云科技股份有限公司 | A face recognition method and device based on multi-task learning |
| CN113850243A (en) * | 2021-11-29 | 2021-12-28 | 北京的卢深视科技有限公司 | Model training, face recognition method, electronic device and storage medium |
| CN114241264A (en) * | 2021-12-17 | 2022-03-25 | 深圳尚米网络技术有限公司 | User discrimination model training method, user discrimination method and related device |
| CN113963237A (en) * | 2021-12-22 | 2022-01-21 | 北京的卢深视科技有限公司 | Model training, mask-wearing state detection method, electronic equipment and storage medium |
| CN114913583A (en) * | 2022-05-17 | 2022-08-16 | 中南大学 | Image type detection method and system based on deep learning twin network |
| CN114913583B (en) * | 2022-05-17 | 2024-08-09 | 中南大学 | Image type detection method and system based on deep learning twin network |
| CN114898454A (en) * | 2022-05-31 | 2022-08-12 | 中国第一汽车股份有限公司 | Neural network model training method and device, electronic equipment and medium |
| CN114898454B (en) * | 2022-05-31 | 2026-01-06 | 中国第一汽车股份有限公司 | A method, apparatus, electronic device, and medium for training a neural network model. |
| CN116503930A (en) * | 2023-05-11 | 2023-07-28 | 上海科江电子信息技术有限公司 | A 3D Face Recognition System with Double Sampling Mechanism and Multi-branch Parallel Training |
| CN116503930B (en) * | 2023-05-11 | 2025-08-26 | 上海科江电子信息技术有限公司 | A 3D face recognition system with dual sampling mechanism and multi-branch parallel training |
| CN116758374A (en) * | 2023-06-27 | 2023-09-15 | 支付宝(杭州)信息技术有限公司 | Image recognition model training method, image recognition method and device |
| CN117058440A (en) * | 2023-07-12 | 2023-11-14 | 济南博观智能科技有限公司 | A safety strap identification method, device, equipment and storage medium |
| CN117557888A (en) * | 2024-01-12 | 2024-02-13 | 清华大学深圳国际研究生院 | Face model training method and face recognition method based on metric learning |
| CN117557888B (en) * | 2024-01-12 | 2024-04-12 | 清华大学深圳国际研究生院 | Face model training method and face recognition method based on metric learning |
| CN118366206A (en) * | 2024-06-19 | 2024-07-19 | 苏州元脑智能科技有限公司 | Face recognition method, face recognition device and face recognition access control system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109815801A (en) | Face identification method and device based on deep learning | |
| CN109800648B (en) | Face detection and recognition method and device based on face key point correction | |
| CN113378632B (en) | Pseudo-label optimization-based unsupervised domain adaptive pedestrian re-identification method | |
| CN111709311B (en) | A pedestrian re-identification method based on multi-scale convolution feature fusion | |
| CN112580590B (en) | Finger vein recognition method based on multi-semantic feature fusion network | |
| US8842883B2 (en) | Global classifier with local adaption for objection detection | |
| CN110188225B (en) | Image retrieval method based on sequencing learning and multivariate loss | |
| CN111274915B (en) | Deep local aggregation descriptor extraction method and system for finger vein image | |
| CN107480261A (en) | One kind is based on deep learning fine granularity facial image method for quickly retrieving | |
| CN111126482A (en) | Remote sensing image automatic classification method based on multi-classifier cascade model | |
| CN112784728A (en) | Multi-granularity clothes changing pedestrian re-identification method based on clothing desensitization network | |
| CN106372624B (en) | Face recognition method and system | |
| CN113076833B (en) | Training method of age identification model, face age identification method and related device | |
| CN110751027B (en) | Pedestrian re-identification method based on deep multi-instance learning | |
| CN111325237B (en) | An image recognition method based on attention interaction mechanism | |
| CN111814705B (en) | Pedestrian re-identification method based on batch blocking shielding network | |
| CN116343135B (en) | A vehicle re-identification method based on post-feature fusion from pure vision | |
| CN111460200B (en) | Image retrieval method, model and construction method based on multi-task deep learning | |
| CN109190521A (en) | A kind of construction method of the human face recognition model of knowledge based purification and application | |
| Cheng et al. | A multiclassification method for iris data based on the Hadamard error correction output code and a convolutional network | |
| CN111815582A (en) | A two-dimensional code region detection method with improved background prior and foreground prior | |
| CN112241680A (en) | Multi-mode identity authentication method based on vein similar image knowledge migration network | |
| CN116935411A (en) | Radical-level ancient character recognition method based on character decomposition and reconstruction | |
| CN115100694B (en) | A fast fingerprint retrieval method based on self-supervised neural network | |
| CN110458064A (en) | Combining data-driven and knowledge-driven low-altitude target detection and recognition methods |
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: 20190528 |
|
| RJ01 | Rejection of invention patent application after publication |