WO2021012494A1

WO2021012494A1 - Deep learning-based face recognition method and apparatus, and computer-readable storage medium

Info

Publication number: WO2021012494A1
Application number: PCT/CN2019/116934
Authority: WO
Inventors: 黄秋凤; 李珊珊
Original assignee: 平安科技（深圳）有限公司
Priority date: 2019-07-19
Filing date: 2019-11-10
Publication date: 2021-01-28
Also published as: CN110516544B; CN110516544A

Abstract

A deep learning-based face recognition method and apparatus and a computer-readable storage medium, which relate to artificial intelligence technology. The method comprises: obtaining face image data from a webpage on the basis of crawler technology, and constituting an original face image set (S1); extracting face features of the original face image set according to a Gabor filter to obtain a face feature set, and performing dimension reduction processing on the face feature set according to a down-sampling technique to form a face feature vector set (S2); inputting the face feature vector set into a pre-constructed convolutional neural network model for training until a loss function value in a convolutional neural network is smaller than a preset threshold, and then exiting training (S3); and receiving a user face picture, inputting the user face picture into the convolutional neural network for face recognition, and outputting a recognition result (S4). The described method may implement efficient and accurate face recognition.

Description

Face recognition method, device and computer readable storage medium based on deep learning

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on July 19, 2019, the application number is 201910658687.0, and the invention title is "Face Recognition Method, Device and Computer-readable Storage Medium Based on Deep Learning". The entire content is incorporated in the application by reference.

Technical field

This application relates to the field of artificial intelligence technology, and in particular to a face recognition method, device and computer-readable storage medium based on Gabor filters and convolutional neural networks.

Background technique

Face recognition is a kind of biometric recognition technology based on human facial feature information. At present, face recognition technology mainly uses cameras and other camera equipment to collect images or video streams containing human faces, and automatically detect faces in the images, and then perform a series of related operations on the detected faces. The process of face recognition is the process of extracting and recognizing features from standard face images. Therefore, the quality of the extracted facial image features directly affects the final recognition accuracy, and the recognition model also plays a vital role in the accuracy of face recognition. However, most of the current feature extraction is mainly based on manual feature extraction. This method is restricted by many factors, and the current recognition models are based on traditional machine learning algorithms. Therefore, in general, the face recognition effect is not ideal and the recognition accuracy is not high.

Summary of the invention

This application provides a face recognition method, device and computer-readable storage medium based on deep learning, the main purpose of which is to accurately identify a person from the face picture or video when a user inputs a face picture or video Face result.

In order to achieve the above objective, a face recognition method based on deep learning provided by this application includes:

Obtain face image data from web pages based on crawler technology to form an original face image set;

Extracting face features of the original face image set according to the Gabor filter to obtain a face feature set, and performing dimensionality reduction processing on the face feature set according to a downsampling technique to form a face feature vector set;

Inputting the face feature vector set into a pre-built convolutional neural network model for training, and exiting the training when the loss function value in the convolutional neural network is less than a preset threshold;

A picture of a user's face is received, and the picture of the user's face is input to the convolutional neural network for face recognition, and the recognition result is output.

Optionally, the web page includes a web page of an ORL face database, a Yale face database, an AR face database, and/or a FERET face database.

Optionally, the extracting the face features of the original face image set according to the Gabor filter to obtain the face feature set includes:

A Gabor filter bank composed of several Gabor filters receives the original face image set;

The Gabor filter bank sequentially performs a first convolution operation with pictures in the original face image set to obtain Gabor features;

The Gabor features obtained by each first convolution operation are combined into a set to obtain the face feature set.

Optionally, the first convolution operation is:

O _y,u,v (x ₁ ,x ₂ )=M(x ₁ ,x ₂ )*φ _y,u,v (z)

Where O _{y, u, v} (x ₁ , x ₂ ) is the Gabor feature, M(x ₁ , x ₂ ) is the pixel value coordinates of the picture in the original face image set, φ _{y, u, v} (z) is the convolution function, z is the convolution operator, y, u, and v represent the three components of the picture, where y is the brightness of the picture, and u, v are the chromaticity of the picture.

Optionally, the convolutional neural network includes a sixteen-layer convolutional layer, a sixteen-layer pooling layer, and a fully connected layer; and the input of the face feature vector set to a pre-built convolutional neural network Training in the network model until the loss function value in the convolutional neural network is less than the preset threshold to exit training, including:

After receiving the face feature vector set, the convolutional neural network inputs the face feature vector set to the sixteen-layer convolutional layer and sixteen-layer pooling layer to perform a second convolution operation and a maximum pooling Input to the fully connected layer after transformation operation;

The fully connected layer is combined with the activation function to calculate the training value, and the training value is input into the loss function of the model training layer, and the loss function calculates the loss value, and the magnitude of the loss value and a preset threshold is judged Relationship, until the loss value is less than the preset threshold, the convolutional neural network exits training.

In addition, in order to achieve the above object, this application also provides a face recognition device based on deep learning. The device includes a memory and a processor. The memory stores a person based on deep learning that can run on the processor. A face recognition program, when the face recognition program based on deep learning is executed by the processor, the following steps are implemented:

In addition, in order to achieve the above object, the present application also provides a computer-readable storage medium that stores a face recognition program based on deep learning, and the face recognition program based on deep learning can be One or more processors execute to implement the steps of the face recognition method based on deep learning as described above.

The face recognition method, device and computer readable storage medium based on deep learning proposed in this application can use crawler technology to collect a large number of high-quality face data sets from the Internet, which is ready for subsequent face feature analysis and recognition Pre-based, and because most faces do not occupy the entire picture or video, according to the shape of the Gabor filter, the features of the face part are extracted from the entire picture or video, which not only reduces the cumbersomeness of manually extracting features, but also At the same time, sufficient preparations are made for the subsequent analysis of the facial features by the convolutional neural network, which can effectively analyze the facial features and produce accurate face recognition effects. Therefore, this application can achieve an efficient and accurate face recognition effect.

Description of the drawings

FIG. 1 is a schematic flowchart of a face recognition method based on deep learning provided by an embodiment of this application;

2 is a Gabor feature generation diagram of a face recognition method based on deep learning provided by an embodiment of this application;

3 is a schematic diagram of the internal structure of a face recognition device based on deep learning provided by an embodiment of the application;

4 is a schematic diagram of modules of a face recognition program based on deep learning in a face recognition device based on deep learning provided by an embodiment of the application.

The realization, functional characteristics, and advantages of the purpose of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings.

Detailed ways

It should be understood that the specific embodiments described here are only used to explain the application, and are not used to limit the application.

This application provides a face recognition method based on deep learning. Referring to FIG. 1, it is a schematic flowchart of a face recognition method based on deep learning provided by an embodiment of this application. The method can be executed by a device, and the device can be implemented by software and/or hardware.

In this embodiment, the face recognition method based on deep learning includes:

S1. Obtain face image data from web pages based on crawler technology, such as web pages in several face image databases, to form an original face image set.

The several face image databases include ORL face database, Yale face database, AR face database, and/or FERET face database, etc. Among them, the Yale face database includes 15 people, including 11 photos per person, and each photo has changes in lighting conditions, changes in facial expressions, etc.; the FERET face database is Counterdrug Technology Transfer Program ( CTTP) In order to promote the further optimization of face recognition technology, a face database collection activity of Face Recognition Technology (Face Recognition Technology, FERET for short) is initiated. The FERET face database includes a general face database and a general test standard. The same face picture includes pictures of different expressions, lighting, postures and age groups.

Preferably, this application uses the Urllib module of python to read web page data, such as reading the web page of the FERET face database, and capture the face image data in the web page of the FERET face database, and combine these The data composes the original face image set. Similarly, the Urllib module reads web pages such as Yale face database, AR face database, etc., and captures the face image data before placing it in the original face image set.

S2. Extract the face features of the original face image set according to the Gabor filter to obtain a face feature set, and perform dimensionality reduction processing on the face feature set according to the downsampling technology to form a face feature vector set.

Preferably, this application composes several Gabor filters into a Gabor filter bank, and after the Gabor filter bank receives the original face image set, the Gabor filter bank is in turn with those in the original face image set. The picture is subjected to the first convolution operation to obtain Gabor features, and the Gabor features obtained from each first convolution operation are combined into a set to obtain the face feature set.

Further, the first convolution operation is:

O _y,u,v (x ₁ ,x ₂ )=M(x ₁ ,x ₂ )*φ _y,u,v (z)

The preferred embodiment of this application selects 40 Gabor filters to form a Gabor filter bank. For example, the 40 Gabor filters form a Gabor filter bank to read an image of the original face image set and compare it with the Gabor filter bank. The filter bank performs the first convolution operation to obtain Gabor features, and the feature dimension of each Gabor feature is 40, and so on, the Gabor features form the face feature set. The change from the original face image to the Gabor feature is shown in Figure 2.

Preferably, the downsampling technology dimensionality reduction processing includes the first feature dimensionality reduction and the second feature dimensionality reduction. The first feature dimensionality reduction is to sequentially extract Gabor features from the face feature set, and based on a sliding window with a matrix dimension of 2*2, from left to right and from top to bottom in the extracted Gabor A mean value sampling with a step length of 2 is performed on the feature, whereby the feature dimension of the extracted Gabor feature is reduced to 1/4 of the original dimension, and the feature dimension becomes 10, and the first feature dimensionality reduction is completed.

Optionally, the feature dimension of the Gabor feature is reduced to 1/4 of the original dimension, and then an RBM model is connected to perform the second feature reduction. The RBM is an energy model (Energy based model, EBM), which is derived from Evolved from the physical energy model, the RBM model receives input data and solves the probability distribution of the input data according to an energy function, and obtains output data after optimization based on the probability distribution. Specifically, the second feature reduction uses the face feature set after the first feature reduction as the input data of the RBM model. Preferably, the feature dimension of the output feature of the RBM model is 5. In general, the dimensionality reduction processing reduces the feature dimension of Gabor features from 40 to 5, and so on to process each Gabor feature and finally compose the output dimensionality reduction feature into a face feature vector set.

S3. Input the face feature vector set into a pre-built convolutional neural network model for training, and exit the training when the loss function value in the convolutional neural network is less than a preset threshold.

Preferably, the pre-built convolutional neural network includes a sixteen-layer convolutional layer, a sixteen-layer pooling layer, and a fully connected layer. After the convolutional neural network receives the face feature vector set, Input the face feature vector set to the sixteen-layer convolutional layer and the sixteen-layer pooling layer to perform a second convolution operation and a maximum pooling operation, and then input to the fully connected layer;

Further, the fully connected layer is combined with the activation function to calculate the training value, and the training value is input into the loss function of the model training layer. The loss function calculates the loss value, and judges the loss value and the preset value. The size relationship of the threshold value, until the loss value is less than the preset threshold value, the convolutional neural network exits training.

The second convolution operation described in the preferred embodiment of the present application is:

Where ω'is the output data, ω is the input data, k is the size of the convolution kernel, s is the stride of the convolution operation, p is the data zero-filling matrix, and the maximum pooling operation is to select a matrix in the matrix The largest value in the data replaces the entire matrix;

The activation function is:

Where y is the training value, and e is an infinite non-recurring decimal.

The loss value T in the preferred embodiment of the present application is:

Wherein, n is the size of the original picture set, y _t is the training value, μ _t is the original picture set, and the preset threshold is generally set at 0.01.

S4. Receive a face picture of the user, and input the face picture of the user into the convolutional neural network for face recognition, and output the recognition result.

The invention also provides a face recognition device based on deep learning. Referring to FIG. 3, it is a schematic diagram of the internal structure of a face recognition device based on deep learning provided by an embodiment of this application.

In this embodiment, the face recognition apparatus 1 based on deep learning may be a PC (Personal Computer, personal computer), or a terminal device such as a smart phone, a tablet computer, or a portable computer, or a server. The face recognition device 1 based on deep learning at least includes a memory 11, a processor 12, a communication bus 13, and a network interface 14.

Wherein, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of the face recognition device 1 based on deep learning, for example, a hard disk of the face recognition device 1 based on deep learning. In other embodiments, the memory 11 may also be an external storage device of the face recognition device 1 based on deep learning, such as a plug-in hard disk equipped on the face recognition device 1 based on deep learning, and a smart media card (Smart Media Card). , SMC), Secure Digital (SD) card, Flash Card, etc. Further, the memory 11 may also include both an internal storage unit of the face recognition apparatus 1 based on deep learning and an external storage device. The memory 11 can be used not only to store application software and various data installed in the face recognition device 1 based on deep learning, such as the code of the face recognition program 01 based on deep learning, etc., but also to temporarily store the output or The data to be output.

In some embodiments, the processor 12 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip, and is used to run the program code or processing stored in the memory 11 Data, such as the face recognition program 01 based on deep learning.

The communication bus 13 is used to realize the connection and communication between these components.

The network interface 14 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface), and is usually used to establish a communication connection between the device 1 and other electronic devices.

Optionally, the device 1 may also include a user interface. The user interface may include a display (Display) and an input unit such as a keyboard (Keyboard). The optional user interface may also include a standard wired interface and a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc. Among them, the display can also be appropriately called a display screen or a display unit, which is used to display the information processed in the face recognition device 1 based on deep learning and to display a visualized user interface.

FIG. 3 only shows the deep learning-based face recognition device 1 with components 11-14 and the deep-learning-based face recognition program 01. Those skilled in the art can understand that the structure shown in FIG. 3 does not constitute The definition of the face recognition device 1 based on deep learning may include fewer or more components than shown, or a combination of certain components, or different component arrangements.

In the embodiment of the device 1 shown in FIG. 3, the memory 11 stores a face recognition program 01 based on deep learning; the processor 12 implements the following steps when executing the face recognition program 01 based on deep learning stored in the memory 11:

Further, the first convolution operation is:

O _y,u,v (x ₁ ,x ₂ )=M(x ₁ ,x ₂ )*φ _y,u,v (z)

Where O _{y, u, v} (x ₁ , x ₂ ) is the Gabor feature, M(x ₁ , x ₂ ) is the pixel value coordinates of the picture in the original face image set, φ _{y, u, v} (z) is the convolution function, z is the convolution operator, y, u, v represent the three components of the picture, where y is the brightness of the picture, and u, v are the chromaticity of the picture.

The activation function is:

Where y is the training value, and e is an infinite non-recurring decimal.

The loss value T in the preferred embodiment of the present application is:

S4. Receive a face picture of the user, and input the face picture of the user into the convolutional neural network for face recognition, and output a recognition result.

Optionally, in other embodiments, the deep learning-based face recognition program can also be divided into one or more modules, and the one or more modules are stored in the memory 11 and are executed by one or more processors ( This embodiment is executed by the processor 12) to complete this application. The module referred to in this application refers to a series of computer program instruction segments that can complete specific functions, and is used to describe how a face recognition program based on deep learning is based on deep learning. The execution process in the face recognition device.

For example, referring to FIG. 4, which is a schematic diagram of the program modules of the face recognition program based on deep learning in an embodiment of the face recognition device based on deep learning of this application, in this embodiment, the face recognition program based on deep learning The recognition program can be divided into a source data receiving module 10, a feature extraction module 20, a model training module 30, and a face recognition result output module 40. Illustratively:

The source data receiving module 10 is used to obtain face image data from web pages based on crawler technology to form an original face image set.

The feature extraction module 20 is configured to: extract the face features of the original face image set according to the Gabor filter to obtain a face feature set, and perform dimensionality reduction processing on the face feature set according to a downsampling technique to form a face feature Vector set.

The model training module 30 is configured to: input the face feature vector set into a pre-built convolutional neural network model for training, and exit training when the loss function value in the convolutional neural network is less than a preset threshold.

The face recognition result output module 40 is configured to receive a face picture of the user, and input the face picture of the user into the convolutional neural network for face recognition, and output the recognition result.

The above-mentioned source data receiving module 10, feature extraction module 20, model training module 30, face recognition result output module 40, and other program modules that implement functions or operation steps when executed are substantially the same as those in the foregoing embodiment, and will not be repeated here.

In addition, an embodiment of the present application also proposes a computer-readable storage medium that stores a face recognition program based on deep learning, and the face recognition program based on deep learning can be used by one or more Each processor executes to achieve the following operations:

Input the face feature vector set into a pre-built convolutional neural network model for training, and exit training when the loss function value in the convolutional neural network is less than a preset threshold;

The specific implementation of the computer-readable storage medium of the present application is basically the same as the foregoing embodiments of the face recognition device and method based on deep learning, and will not be repeated here.

It should be noted that the serial numbers of the above-mentioned embodiments of the present application are only for description, and do not represent the superiority of the embodiments. And the terms "include", "include" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements not only includes those elements, but also includes The other elements listed may also include elements inherent to the process, device, article, or method. If there are no more restrictions, the element defined by the sentence "including a..." does not exclude the existence of other identical elements in the process, device, article or method that includes the element.

Through the description of the above embodiments, those skilled in the art can clearly understand that the method of the above embodiments can be implemented by means of software plus the necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is better.的实施方式。 Based on this understanding, the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as ROM/RAM) as described above. , Magnetic disk, optical disk), including several instructions to make a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) execute the method described in each embodiment of the present application.

The above are only preferred embodiments of this application, and do not limit the scope of this application. Any equivalent structure or equivalent process transformation made using the content of the description and drawings of this application, or directly or indirectly used in other related technical fields , The same reason is included in the scope of patent protection of this application.

Claims

A face recognition method based on deep learning, characterized in that the method includes:

Obtain face image data from web pages based on crawler technology to form an original face image set;

Extracting face features of the original face image set according to the Gabor filter to obtain a face feature set, and performing dimensionality reduction processing on the face feature set according to a downsampling technique to form a face feature vector set;

Inputting the face feature vector set into a pre-built convolutional neural network model for training, and exiting the training when the loss function value in the convolutional neural network is less than a preset threshold;

A picture of a user's face is received, and the picture of the user's face is input to the convolutional neural network for face recognition, and the recognition result is output.
The face recognition method based on deep learning according to claim 1, wherein the web page comprises a web page of an ORL face database, a Yale face database, an AR face database, and/or a FERET face database.
The face recognition method based on deep learning according to claim 1 or 2, wherein said extracting the face features of the original face image set according to the Gabor filter to obtain the face feature set comprises:

A Gabor filter bank composed of several Gabor filters receives the original face image set;

The Gabor filter bank sequentially performs a first convolution operation with pictures in the original face image set to obtain Gabor features;

The Gabor features obtained by each first convolution operation are combined into a set to obtain the face feature set.
The face recognition method based on deep learning of claim 3, wherein the first convolution operation is:

O y,u,v (x 1 ,x 2 )=M(x 1 ,x 2 )*φ y,u,v (z)

Where O y, u, v (x 1 , x 2 ) is the Gabor feature, M(x 1 , x 2 ) is the pixel value coordinates of the picture in the original face image set, φ y, u, v (z) is the convolution function, z is the convolution operator, y, u, and v represent the three components of the picture, where y is the brightness of the picture, and u, v are the chromaticity of the picture.
The face recognition method based on deep learning according to claim 4, wherein the convolutional neural network includes sixteen convolutional layers, sixteen pooling layers, and one fully connected layer; and Inputting the face feature vector set into a pre-built convolutional neural network model for training until the loss function value in the convolutional neural network is less than a preset threshold to exit training, including:

After receiving the face feature vector set, the convolutional neural network inputs the face feature vector set to the sixteen-layer convolutional layer and sixteen-layer pooling layer to perform a second convolution operation and a maximum pooling Input to the fully connected layer after transformation operation;

The fully connected layer is combined with the activation function to calculate the training value, and the training value is input into the loss function of the model training layer, and the loss function calculates the loss value, and the magnitude of the loss value and a preset threshold is judged Relationship, until the loss value is less than the preset threshold, the convolutional neural network exits training.
A face recognition device based on deep learning, characterized in that the device includes a memory and a processor, and a face recognition program based on deep learning that can be run on the processor is stored on the memory. When the face recognition program based on deep learning is executed by the processor, the following steps are implemented:

Obtain face image data from web pages based on crawler technology to form an original face image set;

Extracting face features of the original face image set according to the Gabor filter to obtain a face feature set, and performing dimensionality reduction processing on the face feature set according to a downsampling technique to form a face feature vector set;

Inputting the face feature vector set into a pre-built convolutional neural network model for training, and exiting the training when the loss function value in the convolutional neural network is less than a preset threshold;

Receive the user's face picture, and input the user's face picture into the convolutional neural network for face recognition, and output the recognition result.
The face recognition device based on deep learning according to claim 6, wherein the web page comprises a web page of an ORL face database, a Yale face database, an AR face database, and/or a FERET face database.
The face recognition device based on deep learning according to claim 6 or 7, characterized in that said extracting the face features of the original face image set according to the Gabor filter to obtain the face feature set, include:

A Gabor filter bank composed of several Gabor filters receives the original face image set;

The Gabor filter bank sequentially performs a first convolution operation with pictures in the original face image set to obtain Gabor features;

The Gabor features obtained by each first convolution operation are combined into a set to obtain the face feature set.
The face recognition device based on deep learning of claim 8, wherein the first convolution operation is:

O y,u,v (x 1 ,x 2 )=M(x 1 ,x 2 )*φ y,u,v (z)

Where O y, u, v (x 1 , x 2 ) is the Gabor feature, M(x 1 , x 2 ) is the pixel value coordinates of the picture in the original face image set, φ y, u, v (z) is the convolution function, z is the convolution operator, y, u, and v represent the three components of the picture, where y is the brightness of the picture, and u, v are the chromaticity of the picture.
The face recognition device based on deep learning according to claim 9, wherein the convolutional neural network comprises sixteen convolutional layers, sixteen pooling layers, and one fully connected layer; and Inputting the face feature vector set into a pre-built convolutional neural network model for training until the loss function value in the convolutional neural network is less than a preset threshold to exit training, including:

After receiving the face feature vector set, the convolutional neural network inputs the face feature vector set to the sixteen-layer convolutional layer and sixteen-layer pooling layer to perform a second convolution operation and a maximum pooling Input to the fully connected layer after transformation operation;

The fully connected layer is combined with the activation function to calculate the training value, and the training value is input into the loss function of the model training layer, and the loss function calculates the loss value, and the magnitude of the loss value and a preset threshold is judged Relationship, until the loss value is less than the preset threshold, the convolutional neural network exits training.
A computer-readable storage medium, characterized in that a face recognition program based on deep learning is stored on the computer-readable storage medium, and the face recognition program based on deep learning can be executed by one or more processors To achieve the following steps:

Obtain face image data from web pages based on crawler technology to form an original face image set;

Extracting face features of the original face image set according to the Gabor filter to obtain a face feature set, and performing dimensionality reduction processing on the face feature set according to a downsampling technique to form a face feature vector set;

Input the face feature vector set into a pre-built convolutional neural network model for training, and exit training when the loss function value in the convolutional neural network is less than a preset threshold;

Receive the user's face picture, and input the user's face picture into the convolutional neural network for face recognition, and output the recognition result.
The computer readable storage medium according to claim 11, wherein the web page comprises a web page of an ORL face database, a Yale face database, an AR face database, and/or a FERET face database.
The computer-readable storage medium according to claim 11 or 12, wherein the extracting the face features of the original face image set according to the Gabor filter to obtain the face feature set comprises:

A Gabor filter bank composed of several Gabor filters receives the original face image set;

The Gabor filter bank sequentially performs a first convolution operation with pictures in the original face image set to obtain Gabor features;

The Gabor features obtained by each first convolution operation are combined into a set to obtain the face feature set.
The computer-readable storage medium according to claim 13, wherein the first convolution operation is:

O y,u,v (x 1 ,x 2 )=M(x 1 ,x 2 )*φ y,u,v (z)

Where O y, u, v (x 1 , x 2 ) is the Gabor feature, M(x 1 , x 2 ) is the pixel value coordinates of the picture in the original face image set, φ y, u, v (z) is the convolution function, z is the convolution operator, y, u, and v represent the three components of the picture, where y is the brightness of the picture, and u, v are the chromaticity of the picture.
The computer-readable storage medium of claim 14, wherein the convolutional neural network includes sixteen convolutional layers, sixteen pooling layers, and a fully connected layer; and the convolutional neural network The face feature vector set is input into a pre-built convolutional neural network model for training, and training is exited until the loss function value in the convolutional neural network is less than a preset threshold, including:

After receiving the face feature vector set, the convolutional neural network inputs the face feature vector set to the sixteen-layer convolutional layer and sixteen-layer pooling layer to perform a second convolution operation and a maximum pooling Input to the fully connected layer after transformation operation;

The fully connected layer is combined with the activation function to calculate the training value, and the training value is input into the loss function of the model training layer, and the loss function calculates the loss value, and the magnitude of the loss value and a preset threshold is judged Relationship, until the loss value is less than the preset threshold, the convolutional neural network exits training.
A face recognition system based on deep learning, characterized in that, the face recognition system based on deep learning includes:

The source data receiving module is used to: obtain face image data from web pages based on crawler technology to form an original face image set;

The feature extraction module is used to: extract the face features of the original face image set according to the Gabor filter to obtain a face feature set, and perform dimensionality reduction processing on the face feature set according to downsampling technology to form a face feature vector set ；

The model training module is configured to: input the face feature vector set into a pre-built convolutional neural network model for training, and exit the training when the loss function value in the convolutional neural network is less than a preset threshold;

The face recognition result output module is used to: receive a user face picture, input the user face picture into the convolutional neural network for face recognition, and output the recognition result.
The face recognition system based on deep learning according to claim 16, wherein the web page comprises a web page of an ORL face database, a Yale face database, an AR face database, and/or a FERET face database. .
The face recognition system based on deep learning according to claim 16 or 17, wherein said extracting the face features of the original face image set according to the Gabor filter to obtain the face feature set comprises:

A Gabor filter bank composed of several Gabor filters receives the original face image set;

The Gabor filter bank sequentially performs a first convolution operation with pictures in the original face image set to obtain Gabor features;

The Gabor features obtained by each first convolution operation are combined into a set to obtain the face feature set.
The face recognition system based on deep learning of claim 18, wherein the first convolution operation is:

O y,u,v (x 1 ,x 2 )=M(x 1 ,x 2 )*φ y,u,v (z)

Where O y, u, v (x 1 , x 2 ) is the Gabor feature, M(x 1 , x 2 ) is the pixel value coordinates of the picture in the original face image set, φ y, u, v (z) is the convolution function, z is the convolution operator, y, u, and v represent the three components of the picture, where y is the brightness of the picture, and u, v are the chromaticity of the picture.
The face recognition system based on deep learning of claim 19, wherein the convolutional neural network includes sixteen convolutional layers, sixteen pooling layers, and one fully connected layer; and Inputting the face feature vector set into a pre-built convolutional neural network model for training until the loss function value in the convolutional neural network is less than a preset threshold to exit training, including:

After receiving the face feature vector set, the convolutional neural network inputs the face feature vector set to the sixteen-layer convolutional layer and sixteen-layer pooling layer to perform a second convolution operation and a maximum pooling Input to the fully connected layer after transformation operation;

The fully connected layer is combined with the activation function to calculate the training value, the training value is input into the loss function of the model training layer, the loss function calculates the loss value, and the size of the loss value and a preset threshold is judged Relationship, until the loss value is less than the preset threshold, the convolutional neural network exits training.