An artificial intelligence face recognition method and system

Abstract

The present invention discloses an artificial intelligence face recognition method and system, which relates to the technical field of image processing and mainly solves the problem of relatively low accuracy of existing face recognition. It mainly includes obtaining the first facial image data obtained by photographing the target face at a first angle and the facial adjustment features matching the target face; adjusting the first facial features recognized in the first facial image data based on the facial adjustment features to generate the second facial image data, and retrieving the learning coefficient matching the third facial image data obtained by photographing at a second angle and the second facial image data; training the initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data and the learning coefficient to obtain a facial feature recognition model with completed model training; and performing recognition processing on the collected face images based on the facial feature recognition model to obtain the face recognition result.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and particularly to an artificial intelligence face recognition method and system. Background Art

[0002] With the rapid development of artificial intelligence, all walks of life have begun to vigorously develop artificial intelligence technologies to meet different industrial production or digital service needs. In particular, in order to meet the demand for intelligent access rights, artificial intelligence algorithms are used to develop face recognition technologies.

[0003] Currently, when performing face recognition based on artificial intelligence, the image obtained by photographing a face is usually used as the input of the facial feature recognition model. After running the recognition model, the recognition results of each facial feature are obtained. Among them, when the facial feature recognition model is learning, it is necessary to use the facial images of people with permissions as training samples for learning. However, due to the occurrence of situations such as wearing glasses, masks or makeup on the face during the later face recognition process, the accuracy of face recognition is greatly affected. Summary of the Invention

[0004] In view of this, the present invention provides an artificial intelligence face recognition method and system, mainly aiming at the problem of low accuracy of existing face recognition.

[0005] According to one aspect of the present invention, an artificial intelligence face recognition method is provided, including:

[0006] Obtaining first facial image data obtained by photographing a target face at a first angle and facial adjustment features matching the target face;

[0007] Adjusting the first facial features recognized in the first facial image data based on the facial adjustment features to generate second facial image data, and retrieving a learning coefficient matching the third facial image data obtained by photographing at a second angle and the second facial image data, where the learning coefficient is used to represent the learning intensity of model training;

[0008] Training an initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data and the learning coefficient to obtain a facial feature recognition model that has completed model training;

[0009] Performing recognition processing on the collected face images based on the facial feature recognition model to obtain face recognition results.

[0010] Further, the obtaining first facial image data obtained by photographing a target face at a first angle and facial adjustment features matching the target face includes:

[0011] Display the first angle and obtain first facial image data captured of the target face at the first angle, where the first angle includes at least four shooting angles in different directions;

[0012] Display facial feature adjustment items, and determine facial adjustment features through the adjustment items selected by the target user from the facial feature adjustment items, where the facial feature adjustment items include content items for adjusting different facial features.

[0013] Further, adjusting the first facial features recognized in the first facial image data based on the facial adjustment features to generate second facial image data includes:

[0014] Determine a feature adjustment coefficient of the facial adjustment features, and identify the first facial features in the first facial image data based on a feature segmentation model, where the feature adjustment coefficient includes a pixel size adjustment value and an edge sharpness adjustment value;

[0015] Adjust the first facial features based on the feature adjustment coefficient to obtain the second facial image data.

[0016] Further, retrieving a learning coefficient that matches the third facial image data obtained by shooting at a second angle and the second facial image data includes:

[0017] Obtain third facial image data captured at a second angle, where the second angle is obtained by adjusting a preset angle value based on the first angle;

[0018] Match a learning coefficient that matches the second facial image data and the third facial image data based on a preset learning feature adjustment mapping relationship, where different learning coefficients corresponding to facial feature comparison results are stored in the preset learning feature adjustment mapping relationship.

[0019] Further, matching a learning coefficient that matches the second facial image data and the third facial image data based on a preset learning feature adjustment mapping relationship includes:

[0020] Extract the facial features in the second facial image data and the third facial image data, and compare the facial features according to pixel features and color features to obtain a facial feature comparison result;

[0021] Query a matching learning coefficient from the preset learning feature adjustment mapping relationship based on the facial feature comparison result.

[0022] Further, training the initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data, and the learning coefficient to obtain a facial feature recognition model with completed model training includes:

[0023] Construct an initial facial feature recognition model of a three-layer convolutional neural network. The initial facial feature recognition model is a multi-input multi-output network, and the weights between the convolutional layers are configured based on the learning coefficient;

[0024] Determine the learning correlation parameters between the first facial image data, the second facial image data, and the third facial image data, and determine the model training parameters based on the learning correlation parameters and the learning coefficient;

[0025] Train the initial facial feature recognition model based on the first facial image data, the second facial image data, and the third facial image data with labeled feature recognition labels, and complete the model training to obtain a facial feature recognition model when the model training parameters are matched.

[0026] Further, the method further includes:

[0027] Analyze the facial labels in the face recognition result, and perform caching or counting according to the access rights corresponding to the facial labels.

[0028] According to another aspect of the present invention, an artificial intelligence face recognition system is provided, including:

[0029] An acquisition module for acquiring first facial image data obtained by photographing a target face at a first angle and facial adjustment features matching the target face;

[0030] An adjustment module for adjusting the first facial features recognized in the first facial image data based on the facial adjustment features to generate second facial image data, and retrieving a learning coefficient matching the third facial image data obtained by photographing at a second angle and the second facial image data. The learning coefficient is used to represent the learning intensity of model training;

[0031] A training module for training an initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data, and the learning coefficient to obtain a facial feature recognition model with completed model training;

[0032] A recognition module for performing recognition processing on the acquired face image based on the facial feature recognition model to obtain a face recognition result.

[0033] Further, the obtaining module includes:

[0034] A first obtaining unit, configured to display a first angle and obtain first facial image data obtained by photographing a target face according to the first angle, where the first angle includes at least four shooting angles in different directions;

[0035] A first determining unit, configured to display facial feature adjustment items and determine facial adjustment features through the adjustment items selected by the target user from the facial feature adjustment items, where the facial feature adjustment items include content items required for adjusting different facial features.

[0036] Further, the retrieving module includes:

[0037] An identifying unit, configured to determine a feature adjustment coefficient of the facial adjustment features and identify first facial features in the first facial image data based on a feature segmentation model, where the feature adjustment coefficient includes a pixel size adjustment value and an edge sharpness adjustment value;

[0038] An adjusting unit, configured to adjust the first facial features based on the feature adjustment coefficient to obtain the second facial image data.

[0039] Further, the retrieving module further includes:

[0040] A second obtaining unit, configured to obtain third facial image data obtained by photographing according to a second angle, where the second angle is obtained by adjusting a preset angle value based on the first angle;

[0041] A matching unit, configured to match a learning coefficient that matches the second facial image data and the third facial image data based on a preset learning feature adjustment mapping relationship, where different learning coefficients corresponding to facial feature comparison results are stored in the preset learning feature adjustment mapping relationship.

[0042] Further, in a specific application scenario, the matching unit is specifically configured to extract facial features in the second facial image data and the third facial image data, compare the facial features according to pixel features and color features to obtain a facial feature comparison result; and query a matching learning coefficient from the preset learning feature adjustment mapping relationship based on the facial feature comparison result.

[0043] Further, the training module includes:

[0044] A constructing unit, configured to construct an initial facial feature recognition model of a three-layer convolutional neural network, where the initial facial feature recognition model is a multi-input multi-output network, and the weights between convolutional layers are configured based on the learning coefficient;

[0045] A second determination unit, configured to determine learning correlation parameters among the first facial image data, the second facial image data, and the third facial image data, and determine model training parameters based on the learning correlation parameters and the learning coefficient;

[0046] A training unit, configured to train the initial facial feature recognition model based on the first facial image data, the second facial image data, and the third facial image data for identifying the labeled feature, and complete the model training to obtain a facial feature recognition model when the model training parameters are matched.

[0047] Further, the system further includes:

[0048] A cache counting module, configured to parse the facial labels in the face recognition result, and perform caching or counting according to the access permissions corresponding to the facial labels.

[0049] According to another aspect of the present invention, there is provided a storage medium storing at least one executable instruction, and the executable instruction causes a processor to perform operations corresponding to the above-mentioned artificial intelligence face recognition method.

[0050] According to still another aspect of the present invention, there is provided a terminal including: a processor, a memory, a communication interface, and a communication bus, and the processor, the memory, and the communication interface complete communication with each other through the communication bus;

[0051] The memory is used to store at least one executable instruction, and the executable instruction causes the processor to perform operations corresponding to the above-mentioned artificial intelligence face recognition method.

[0052] By means of the above technical solutions, the technical solutions provided by the embodiments of the present invention have at least the following advantages:

[0053] The present invention provides an artificial intelligence face recognition method and system. First, obtain the first facial image data captured at a first angle of a target face and the facial adjustment features matching the target face; adjust the first facial features recognized in the first facial image data based on the facial adjustment features to generate the second facial image data, and retrieve the learning coefficient matching the third facial image data captured at a second angle and the second facial image data, where the learning coefficient is used to represent the learning intensity of model training; train an initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data, and the learning coefficient to obtain a facial feature recognition model with completed model training; perform recognition processing on the captured face images based on the facial feature recognition model to obtain a face recognition result. Compared with the prior art, in the embodiment of the present invention, the facial features of the first facial image data are adjusted through the facial adjustment features, and the model learning intensity is determined based on the second facial image data obtained by the facial feature adjustment and the third facial image data captured at different angles, improving the comprehensiveness and diversity of the face image samples, greatly improving the accuracy of face recognition. At the same time, the matching degree between model training and sample features is greatly increased, achieving an equilibrium configuration between the accuracy and efficiency of model training, thereby greatly improving the accuracy of the facial feature recognition model.

[0054] The above description is only an overview of the technical solution of the present invention. In order to be able to understand the technical means of the present invention more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of the present invention more obvious and understandable, the following specifically illustrates the specific embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. And throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0056] Figure 1 Shows a flowchart of an artificial intelligence face recognition method provided by an embodiment of the present invention;

[0057] Figure 2 Shows a flowchart of another artificial intelligence face recognition method provided by an embodiment of the present invention;

[0058] Figure 3 Shows a block diagram of the composition of an artificial intelligence face recognition system provided by an embodiment of the present invention;

[0059] Figure 4The figure shows a schematic structural diagram of a terminal provided by an embodiment of the present invention. Detailed implementation manners

[0060] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0061] An embodiment of the present invention provides an artificial intelligence face recognition method, as Figure 1 shown, the method includes:

[0062] 101. Obtain first facial image data obtained by photographing a target face at a first angle and facial adjustment features matching the target face.

[0063] In an embodiment of the present invention, during the process of sampling the image of the target face, it is necessary to photograph at a preset first angle. Among them, the target face is the face of the target user who is expected to perform face recognition. The first angle may include multiple shooting directions, for example, front, left, right, upward view, downward view, etc. In addition to inputting the face image, the target user can also input his own facial adjustment features through a human-computer interaction interface or a data input interface. Among them, the facial adjustment features are the additional facial features generated after operations such as beautifying, covering, and wearing on the face. For example, applying makeup, wearing a mask, wearing glasses, etc. The user can select one or more of the facial adjustment features according to his daily dressing habits, and after the user makes a selection, the system will associate the selected facial adjustment features with the target face so as to retrieve the facial adjustment features matching the target face.

[0064] It should be noted that by providing the configuration function of facial adjustment features to the user, the content of the facial image data is enriched, and at the same time, the customized input of the facial image data is realized, providing more accurate and targeted training samples for the training of the subsequent initial facial feature recognition model, thereby improving the learning ability and robustness of the model.

[0065] 102. Adjust the first facial features recognized in the first facial image data based on the facial adjustment features to generate second facial image data, and retrieve the learning coefficients matching the third facial image data obtained by photographing at a second angle and the second facial image data.

[0066] In an embodiment of the present invention, after obtaining the first facial image data and facial adjustment features of a target face, each facial feature in the first facial image data is respectively recognized to obtain the first facial features, and the matching facial features in the first facial features are adjusted based on the facial adjustment features to generate the second facial image data. For example, if the facial adjustment feature is glasses, the eyes in the first facial features are adjusted using the glasses; if the facial adjustment feature is a mask, the nose, mouth, and chin in the first facial features are adjusted using the mask. Furthermore, the learning coefficient is retrieved based on the third facial image and the second facial image. The third facial image is a facial image taken according to a second angle. The second angle is a shooting direction different from the first angle. For example, the first angle includes the front, left, and right, and the second angle is the 45° oblique direction to the left and the 45° oblique direction to the right. The learning coefficient can be matched through a preset mapping relationship between the learning coefficient and the feature similarity degree between the above two images. For example, the mapping relationship between the learning coefficient and the feature similarity of the third facial image data and the second facial image data, the mapping relationship between the learning coefficient and the feature matching result of the third facial image data and the second facial image data, etc.

[0067] It should be noted that the learning coefficient is used to represent the learning intensity of model training, and the learning intensity can include one or more of the learning rate, batch size, number of iterations, and number of loops. If the similarity between the adjusted second facial image data and the unadjusted third facial image data is high, the number of iterations and the number of loops can be reduced, and the batch size can be increased; if the similarity between the adjusted second facial image data and the unadjusted third facial image data is low, the number of iterations and the number of loops can be increased, and the batch size can be increased. Determining the learning coefficient based on the second facial image data and the third facial image data can make the model training parameters corresponding to different target faces different, thereby improving the flexibility and pertinence of model training.

[0068] 103. Train the initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data, and the learning coefficient to obtain a facial feature recognition model that has completed model training.

[0069] In an embodiment of the present invention, the first facial image data, the second facial image data, and the third facial image data are used as training samples, and the learning coefficient is used as a parameter in the model training process to train the initial facial feature recognition model, thereby obtaining a facial feature recognition model that has completed training. Among them, the initial facial feature recognition model can be a convolutional neural network, a recurrent neural network, or a support vector machine. Specifically, it can be a pre-trained model such as VGGFace, ResNet, Inception, FaceNet, etc., or an improved model of the above models. The embodiments of the present invention do not make specific limitations.

[0070] Determine the learning intensity of model training based on the comparison result between the second facial image data adjusted by the facial adjustment feature and the collected third facial image data, and use the first facial image data, the second facial image data, and the third facial image data as training samples to train the initial facial feature recognition model, which can improve the speed of model training as much as possible while ensuring the accuracy of the model, thereby improving the model training efficiency.

[0071] 104. Perform recognition processing on the collected face image based on the facial feature recognition model to obtain a face recognition result.

[0072] In the embodiment of the present invention, after the training of the initial facial feature recognition model is completed. When the user makes a face recognition request through the client, the image acquisition terminal acquires the face image of the target user, and recognizes the face image through the facial feature recognition model to obtain the face recognition result of the current target user. Among them, the face recognition result includes the recognition result of the face image corresponding to the corresponding identity or permission, and the matching result with other relevant information. For example, the identity of the target user corresponding to the face image, the permission of the target user corresponding to the face image, and whether the identity of the target user corresponding to the face image matches the card swiping information, etc.

[0073] It should be noted that the training samples of the facial feature recognition model include face images of the target user from multiple angles, ensuring the comprehensiveness of the image angles, reducing the angle requirements of the face images, and improving the efficiency of face image recognition. Further, the adjusted image of the originally collected face image according to the facial adjustment feature selected by the target user ensures the learning ability of the model, and can also ensure the accuracy of face recognition in the scenario where the target user has makeup or wears something, thereby improving the efficiency and accuracy of face recognition.

[0074] In an embodiment of the present invention, for further illustration and limitation, as Figure 2 shown, step 101 of obtaining the first facial image data obtained by photographing the target face at the first angle and the facial adjustment feature matching the target face includes:

[0075] 201. Display the first angle, and obtain the first facial image data obtained by photographing the target face according to the first angle.

[0076] 202. Display the facial feature adjustment items, and determine the facial adjustment feature through the adjustment items selected by the target user in the facial feature adjustment items.

[0077] In the embodiments of the present invention, the client for collecting images of the target user is a human-computer interaction device with a display function and an input function. When the user initiates a face image collection request, the first angle is displayed through the human-computer interaction interface, so that the user can take pictures according to the specific shooting angles defined in the first angle. Among them, the first angle includes at least four shooting angles in different directions. For example, frontal shooting, direct left-side shooting, direct right-side shooting, direct downward shooting, etc. The first facial image data also includes images obtained by shooting from at least four directions. After the face image collection of the first angle is completed, facial feature adjustment items are output on the display interface for the target user to select the required facial feature adjustment items. Among them, the facial feature adjustment items include the content items that need to be adjusted for different facial features. For example, applying makeup, wearing a hat, wearing glasses, wearing a mask, etc. Among them, applying makeup can be further divided into light makeup and heavy makeup, and glasses can be divided into sunglasses and ordinary myopia glasses, etc. The user can select according to their daily makeup and dressing habits. For example, if the current user usually wears heavy makeup and a mask, they can select heavy makeup and a hat from the facial feature adjustment items on the interface; if the current user often wears myopia glasses and light makeup, they can select light makeup and glasses from the facial feature adjustment items on the interface. When the target user completes the selection of the facial feature adjustment items and clicks the control for indicating confirmation or submission, a facial adjustment feature associated with this user is generated according to the user's selection operation.

[0078] It should be noted that during the process of collecting face images of the target user, the facial feature adjustment items are displayed to the user through the human-computer interaction interface, so that the user can select the adjustment features of the image according to their personal daily habits, realizing the customized configuration of the facial adjustment features, avoiding adjusting all the features of the face image, greatly reducing the computational amount of image feature processing and the training difficulty of the facial feature recognition model, making the training of the model more targeted, and thus improving the accuracy of model recognition. In addition, training the facial feature recognition model based on the image after feature adjustment enables the model to have accurate recognition ability for users under different disguises, avoiding the situation where the user fails to recognize multiple times, and thus greatly improving the efficiency of face recognition.

[0079] In an embodiment of the present invention, for further explanation and limitation, the step of adjusting the first facial feature recognized in the first facial image data based on the facial adjustment feature to generate the second facial image data includes:

[0080] Determine the feature adjustment coefficient of the facial adjustment feature, and identify the first facial feature in the first facial image data based on the feature segmentation model;

[0081] Adjust the first facial feature based on the feature adjustment coefficient to obtain the second facial image data.

[0082] In an embodiment of the present invention, the feature adjustment coefficient is used to characterize the amplitude of feature adjustment. The larger the value of the feature adjustment coefficient, the greater the adjustment amplitude. The feature adjustment coefficient includes a pixel size adjustment value and an edge sharpness adjustment value. Among them, the edge sharpness adjustment value can be further divided into a sharpness adjustment value, a detail adjustment value, a contour adjustment value, etc. The feature adjustment coefficient varies according to different facial adjustment features. For example, the feature adjustment coefficient for light makeup is less than that for heavy makeup, and the feature adjustment coefficient for glasses is less than that for a mask. The determination of the facial adjustment coefficient can be based on a pre-constructed mapping relationship between different facial adjustment features and different adjustment coefficients. Facial adjustment features are adjusted separately for each feature in the face image. Therefore, before adjustment based on the feature adjustment coefficient, it is also necessary to perform feature segmentation and recognition on the first facial data through a feature segmentation model to obtain multiple facial features that make up the face image, that is, the first facial features. Among them, the feature segmentation model can be a fully convolutional network, a U-Net, or a segmentation model obtained by combining a deep convolutional neural network and a probability graph model. The embodiment of the present invention does not make specific limitations. The first facial features can include multiple features such as the nose, forehead, eyes, cheeks, mouth, etc. When the facial adjustment features include two or more, the adjustment of the facial image includes both the adjustment of the corresponding first facial features based on a single facial adjustment feature and the adjustment of two or more features in the first facial features based on the permutation and combination results of the facial adjustment features. For example, if the facial adjustment features include makeup, glasses, and a mask, the second facial image data includes images adjusted based on makeup, images adjusted based on glasses, images adjusted based on a mask, as well as images adjusted based on makeup and glasses, images adjusted based on a mask and makeup, images adjusted based on a mask and glasses, and images adjusted based on glasses, mask, and makeup together. Among them, the specific feature adjustment process can be adjusted in a manner similar to the function of a beauty camera through a preset beauty template. For example, an eye makeup template, a skin tone adjustment template, a glasses template, a hat template, etc.

[0083] In an embodiment of the present invention, for further illustration and limitation, the steps of retrieving the learning coefficient that matches the third facial image data obtained by shooting at the second angle and the second facial image data include:

[0084] Obtain the third facial image data obtained by shooting at the second angle;

[0085] Based on the preset learning feature adjustment mapping relationship, match the learning coefficient that matches the second facial image data and the third facial image data.

[0086] In an embodiment of the present invention, in order to collect more comprehensive face image information, on the basis of sampling face images at a first angle, it is also necessary to collect third facial image data based on a second angle. The second angle is obtained by adjusting a preset angle value on the basis of the first angle. For example, if the pre-trial angle value is 45°, and the first angles include frontal shooting, direct left shooting, direct right shooting, and direct downward shooting, then the second angles include 45° to the left and 45° to the right with the front as the reference for the shooting direction, and 45° upward with the direct downward as the reference for the shooting angle. After collecting the third facial image data, compare the second facial image data with the third facial image data to obtain a facial feature comparison result between the two groups of images, and then match a learning coefficient that matches this facial feature comparison result from the preset learning feature adjustment mapping relationship. Different learning coefficients corresponding to different facial feature comparison results are stored in the preset learning feature adjustment mapping relationship.

[0087] It should be noted that since the feature adjustment coefficients and adjustment ranges corresponding to different facial feature adjustment items are different, the differences in the face images of different target users after the same facial feature adjustment are also different, and the difficulty levels of model training are also different. Therefore, compare the face images after facial feature adjustment with the face images at different angles to determine the degree of change in the face images after feature adjustment, and then determine the training intensity of the model according to the different degrees of change. If the degree of change is large, the training intensity of the model is correspondingly large; if the degree of change is small, the training intensity of the model is correspondingly small. By determining the learning coefficient, the training process of facial feature recognition can be more accurately controlled, the learning ability of the model can be ensured, and the training difficulty and the number of cycles can be reduced as much as possible, thereby improving the efficiency of model training.

[0088] In an embodiment of the present invention, for further illustration and limitation, the step of matching a learning coefficient that matches the second facial image data and the third facial image data based on the preset learning feature adjustment mapping relationship includes:

[0089] Extract the facial features in the second facial image data and the third facial image data, and compare the facial features according to pixel features and color features to obtain a facial feature comparison result;

[0090] Query a matching learning coefficient from the preset learning feature adjustment mapping relationship based on the facial feature comparison result.

[0091] In the embodiments of the present invention, in order to ensure the accuracy of the comparison between the second facial image data and the third facial image data, facial features are extracted from the images respectively, and each facial feature is compared from two dimensions of pixel features and color features. Among them, facial features can be extracted by SIFT (Scale-Invariant Feature Transform) or HOG (Histogram of Oriented Gradients) algorithms. Furthermore, the facial features at the same position in the second facial image data and the third facial image data are compared for similarity according to color features and pixel features respectively. For a certain facial feature, the comparison results of its color feature and pixel feature respectively can be added as the comparison result of this facial feature, or the comparison result of the color feature or pixel feature with the highest similarity can be used as the comparison result of this facial feature. The embodiments of the present invention do not make specific limitations. After obtaining the comparison results of each facial feature in the facial image, a learning coefficient that matches the current facial feature comparison result is matched from the preset learning feature adjustment mapping relationship. Among them, the preset learning feature adjustment mapping relationship stores the learning coefficients corresponding to different facial feature comparison results.

[0092] In an embodiment of the present invention, for further illustration and limitation, the steps of training the initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data, and the learning coefficient to obtain the facial feature recognition model after completing the model training include:

[0093] Construct an initial facial feature recognition model of a three-layer convolutional neural network;

[0094] Determine the learning correlation parameters among the first facial image data, the second facial image data, and the third facial image data, and determine the model training parameters based on the learning correlation parameters and the learning coefficient;

[0095] Train the initial facial feature recognition model based on the first facial image data, the second facial image data, and the third facial image data with labeled feature recognition labels, and complete the model training to obtain the facial feature recognition model when the model training parameters are matched.

[0096] In the embodiments of the present invention, the initial facial feature recognition model is a three-layer convolutional neural network of a multi-input multi-output network. The weights between each convolutional layer are configured based on the learning coefficient. Among them, different convolutional layers correspond to the extraction of different facial features. Since different facial feature comparison results determine the corresponding learning coefficients, the weights of the convolutional layers of the corresponding facial features can be configured based on the learning coefficients of the corresponding facial features. Specifically, the weights between convolutional layers can be calculated according to the following formula:

[0097]

[0098] wherein, m i is the learning coefficient corresponding to the facial feature i, n is the total number of facial features, and W is the weight of the convolutional layer corresponding to the facial feature i. is an adjustment coefficient, which can be a number less than 1 such as 0.3 or 0.5. The learning correlation parameter is determined based on the similarity matching results among the first facial image data, the second facial image data, and the third facial image data. The greater the similarity, the greater the learning correlation parameter, and the relatively smaller the allowable error for training; the smaller the similarity, the smaller the learning correlation parameter, and the relatively larger the allowable error for training. For example, if the average similarity of the three facial images is 80%, the learning correlation parameter is 0.8; if the average similarity of the three facial images is 50%, the learning correlation parameter is 0.5. Further, the model training parameters are determined according to the learning coefficient and the learning correlation parameter. Among them, the model training parameters include the learning rate, batch size, number of iterations, number of loops, and target error. Among them, the learning rate, batch size, number of iterations, and number of loops are determined according to the mapping relationship between different learning coefficients and different learning rates, different batch sizes, different numbers of iterations, and different numbers of loops. The target error is calculated based on the learning correlation parameter and the learning coefficient. The specific formula is:

[0099]

[0100] wherein, θ is the base error, m i is the learning coefficient corresponding to the facial feature i, n is the total number of facial features, k is the learning correlation parameter, and μ is the target error. Each feature in the first facial image data, the second facial image data, and the third facial image data is respectively marked, and the initial facial feature recognition model is trained based on the marked first facial image data, second facial image data, and third facial image data. When the real-time training parameters of the model satisfy the model training parameters, the model training is completed, and a trained facial feature recognition model is obtained. Among them, it can be determined that the model training is completed when both the number of iterations and the learning rate reach a certain value; it can also be determined that the model training is completed when the error value is less than or equal to the target error.

[0101] Based on the learning correlation parameter and the learning coefficient, the number of iterations and the error of the model training are jointly determined, so that the training accuracy of the model can be adjusted according to the degree of difference between the features and the images, and on the basis of ensuring the training accuracy of the model, the training efficiency of the model is improved as much as possible.

[0102] In an embodiment of the present invention, for further illustration and limitation, the method further includes:

[0103] Parse the face labels in the face recognition result, and perform caching or counting according to the access rights corresponding to the face labels.

[0104] In the embodiments of the present invention, the above-mentioned trained facial feature recognition model can be specifically applied to multiple application scenarios such as factories, scientific research institutions, schools, and enterprises. In different application scenarios, the content of the target label and the access rights can be customized. For example, the face label can be information such as the name, employee number, ID number, and student number of the target user. The access rights can be punching in and out for work, entering and leaving the factory area during working hours, entering and leaving the campus on weekdays, entering and leaving important departments, entering and leaving key control areas, etc. For the same face label, each time the recognition of the corresponding image is completed, that is, when the recognition result of the face image is the same face label, the face recognition information can be cached or the number of face recognition times can be counted under the corresponding access rights. Thus, the statistics and records of different personnel and the occurrence time of different access rights can be realized, so as to statistically manage and control the behavior of personnel. For example, if the current access right is punching in for work, the current successful punching time is cached, and the number of times of punching in for work of the recognized face label is incremented by 1.

[0105] The present invention provides an artificial intelligence face recognition method. First, obtain the first facial image data obtained by photographing the target face at a first angle and the facial adjustment features matching the target face; adjust the first facial features recognized in the first facial image data based on the facial adjustment features to generate second facial image data, and retrieve the learning coefficient matching the third facial image data obtained by photographing at a second angle and the second facial image data, where the learning coefficient is used to represent the learning intensity of model training; train the initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data, and the learning coefficient to obtain a facial feature recognition model that has completed model training; perform recognition processing on the collected face images based on the facial feature recognition model to obtain a face recognition result. Compared with the prior art, in the embodiments of the present invention, the facial features of the first facial image data are adjusted through the facial adjustment features, improving the comprehensiveness and diversity of the face image samples, greatly improving the accuracy of face recognition. In addition, the model learning intensity is determined based on the second facial image data obtained by facial feature adjustment and the third facial image data collected at different angles, greatly increasing the matching degree between model training and sample features, realizing the balanced configuration between model training accuracy and efficiency, and thus greatly improving the accuracy of the facial feature recognition model.

[0106] Further, as for the above Figure 1For the implementation of the method described above, an embodiment of the present invention provides an artificial intelligence face recognition system, as Figure 3 shown, the system includes:

[0107] An acquisition module 31, configured to acquire first facial image data obtained by photographing a target face at a first angle and facial adjustment features matching the target face;

[0108] An extraction module 32, configured to adjust the first facial features recognized in the first facial image data based on the facial adjustment features to generate second facial image data, and extract a learning coefficient matching the third facial image data obtained by photographing at a second angle and the second facial image data, where the learning coefficient is used to characterize the learning intensity of model training;

[0109] A training module 33, configured to train an initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data, and the learning coefficient to obtain a facial feature recognition model that has completed model training;

[0110] An identification module 34, configured to perform identification processing on the acquired face image based on the facial feature recognition model to obtain a face recognition result.

[0111] Further, the acquisition module 31 includes:

[0112] A first acquisition unit, configured to display the first angle and acquire first facial image data obtained by photographing the target face according to the first angle, where the first angle includes at least four shooting angles in different directions;

[0113] A first determination unit, configured to display facial feature adjustment items and determine facial adjustment features through the adjustment items selected by the target user in the facial feature adjustment items, where the facial feature adjustment items include content items for adjusting different facial features.

[0114] Further, the extraction module 32 includes:

[0115] An identification unit, configured to determine a feature adjustment coefficient of the facial adjustment features and identify the first facial features in the first facial image data based on a feature segmentation model, where the feature adjustment coefficient includes a pixel size adjustment value and an edge sharpness adjustment value;

[0116] An adjustment unit, configured to adjust the first facial features based on the feature adjustment coefficient to obtain the second facial image data.

[0117] Further, the extraction module 32 further includes:

[0118] A second acquisition unit, configured to acquire third facial image data captured at a second angle, where the second angle is obtained by adjusting a preset angle value based on the first angle;

[0119] A matching unit, configured to match a learning coefficient that matches the second facial image data and the third facial image data based on a preset learning feature adjustment mapping relationship, where different learning coefficients corresponding to facial feature comparison results are stored in the preset learning feature adjustment mapping relationship.

[0120] Further, in a specific application scenario, the matching unit is specifically configured to extract facial features from the second facial image data and the third facial image data, compare the facial features according to pixel features and color features to obtain a facial feature comparison result; and query a matching learning coefficient from the preset learning feature adjustment mapping relationship based on the facial feature comparison result.

[0121] Further, the training module 33 includes:

[0122] A construction unit, configured to construct an initial facial feature recognition model of a three-layer convolutional neural network, where the initial facial feature recognition model is a multi-input multi-output network, and the weights between convolutional layers are configured based on the learning coefficient;

[0123] A second determination unit, configured to determine a learning correlation parameter between the first facial image data, the second facial image data, and the third facial image data, and determine a model training parameter based on the learning correlation parameter and the learning coefficient;

[0124] A training unit, configured to train the initial facial feature recognition model based on the first facial image data, the second facial image data, and the third facial image data with labeled feature recognition labels, and complete model training to obtain a facial feature recognition model when the model training parameter is matched.

[0125] Further, the system further includes:

[0126] A cache counting module, configured to parse a facial label in the face recognition result and perform caching or counting according to the access permission corresponding to the facial label.

[0127] The present invention provides an artificial intelligence face recognition system. First, first facial image data obtained by photographing a target face at a first angle and facial adjustment features matching the target face are acquired; based on the facial adjustment features, the first facial features recognized in the first facial image data are adjusted to generate second facial image data, and a learning coefficient matching the third facial image data obtained by photographing at a second angle and the second facial image data is retrieved, where the learning coefficient is used to represent the learning intensity of model training; based on the first facial image data, the second facial image data, the third facial image data, and the learning coefficient, an initial facial feature recognition model is trained to obtain a facial feature recognition model with completed model training; and based on the facial feature recognition model, the acquired face image is recognized to obtain a face recognition result. Compared with the prior art, in the embodiment of the present invention, the facial features of the first facial image data are adjusted through facial adjustment features, and the model learning intensity is determined based on the second facial image data obtained by facial feature adjustment and the third facial image data acquired at different angles, improving the comprehensiveness and diversity of face image samples, greatly improving the accuracy of face recognition. At the same time, the matching degree between model training and sample features is greatly increased, achieving an equilibrium configuration between model training accuracy and efficiency, thereby greatly improving the accuracy of the facial feature recognition model.

[0128] According to an embodiment of the present invention, a storage medium is provided, and the storage medium stores at least one executable instruction, and the computer executable instruction can execute the artificial intelligence face recognition method in any of the above method embodiments.

[0129] Figure 4 The structural schematic diagram of a terminal provided according to an embodiment of the present invention is shown. The specific implementation of the terminal is not limited in the specific embodiments of the present invention.

[0130] As Figure 4 shown, the terminal may include: a processor 402, a communication interface 404, a memory 406, and a communication bus 408.

[0131] Among them: the processor 402, the communication interface 404, and the memory 406 communicate with each other through the communication bus 408.

[0132] The communication interface 404 is used for network communication with other devices such as clients or other servers.

[0133] The processor 402 is used to execute the program 410, and specifically can execute the relevant steps in the above embodiments of the artificial intelligence face recognition method.

[0134] Specifically, the program 410 may include program code, which includes computer operation instructions.

[0135] The processor 402 may be a central processing unit (CPU), or a specific integrated circuit (ASIC) (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present invention. One or more processors included in the terminal may be of the same type of processor, such as one or more CPUs; or may be of different types of processors, such as one or more CPUs and one or more ASICs.

[0136] The memory 406 is used to store the program 410. The memory 406 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk memory.

[0137] The program 410 may specifically be used to cause the processor 402 to perform the following operations:

[0138] Obtain first facial image data obtained by capturing the target face at a first angle and facial adjustment features matching the target face;

[0139] Adjust the first facial features recognized in the first facial image data based on the facial adjustment features to generate second facial image data, and retrieve a learning coefficient that matches the third facial image data obtained by capturing at a second angle and the second facial image data, where the learning coefficient is used to characterize the learning intensity of model training;

[0140] Train an initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data, and the learning coefficient to obtain a facial feature recognition model that has completed model training;

[0141] Perform recognition processing on the captured face image based on the facial feature recognition model to obtain a face recognition result.

[0142] Obviously, those skilled in the art should understand that the various modules or steps of the present invention described above can be implemented by a general-purpose computing system. They can be concentrated on a single computing system or distributed over a network composed of multiple computing systems. Optionally, they can be implemented by program codes executable by the computing system. Thus, they can be stored in the storage system and executed by the computing system. And in some cases, the steps shown or described can be executed in a sequence different from that here, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module for implementation. In this way, the present invention is not limited to any specific combination of hardware and software.

[0143] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An artificial intelligence face recognition method, characterized in that: include: Acquire first facial image data obtained by photographing a target face at a first angle and facial adjustment features matching the target face; Among them, the facial adjustment features are additional facial features generated after the face is beautified, covered or worn; Adjusting the first facial feature identified in the first facial image data based on the facial adjustment feature to generate second facial image data, and retrieving a learning coefficient that matches the third facial image data obtained by shooting at a second angle and the second facial image data, wherein the learning coefficient is used to characterize the learning intensity of the model training; Training an initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data, and the learning coefficient to obtain a facial feature recognition model that has completed model training; Performing recognition processing on the collected face image based on the facial feature recognition model to obtain a face recognition result; The step of acquiring first facial image data obtained by photographing a target face at a first angle and facial adjustment features matching the target face comprises: Displaying a first angle, and acquiring first facial image data obtained by photographing a target face at the first angle, wherein the first angle includes at least four directions of shooting angles; Displaying facial feature adjustment items, and determining facial adjustment features through adjustment items selected by the target user from the facial feature adjustment items, wherein the facial feature adjustment items include content items that need to be adjusted for different facial features; and retrieving a learning coefficient that matches the third facial image data obtained by shooting at a second angle and the second facial image data includes: Acquire third facial image data captured at a second angle, where the second angle is obtained by adjusting a preset angle value based on the first angle; adjusting a mapping relationship based on a preset learning feature to match the learning coefficients that match the second facial image data and the third facial image data, wherein the preset learning feature adjustment mapping relationship stores learning coefficients corresponding to different facial feature comparison results; The learning coefficients for adjusting the mapping relationship based on the preset learning features to match the second facial image data and the third facial image data include: extracting facial features from the second facial image data and the third facial image data, and comparing the facial features according to pixel features and color features to obtain a facial feature comparison result; Based on the facial feature comparison result, a matching learning coefficient is queried from the preset learning feature adjustment mapping relationship.

2. The method according to claim 1, characterized in that The step of adjusting the first facial feature identified in the first facial image data based on the facial adjustment feature to generate the second facial image data comprises: Determining a feature adjustment coefficient of the facial adjustment feature, and identifying a first facial feature in the first facial image data based on a feature segmentation model, the feature adjustment coefficient comprising a pixel size adjustment value and an edge definition adjustment value; The first facial feature is adjusted based on the feature adjustment coefficient to obtain the second facial image data.

3. The method according to claim 1, characterized in that The training of the initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data and the learning coefficient to obtain the facial feature recognition model after model training includes: constructing an initial facial feature recognition model of a three-layer convolutional neural network, wherein the initial facial feature recognition model is a multi-input multi-output network, and the weights between the convolutional layers are configured based on the learning coefficient; determining a learning association parameter between the first facial image data, the second facial image data, and the third facial image data, and determining a model training parameter based on the learning association parameter and the learning coefficient; The initial facial feature recognition model is trained based on the first facial image data, the second facial image data, and the third facial image data marked with feature recognition tags, and when the model training parameters are matched, the model training is completed to obtain a facial feature recognition model.

4. The method according to any one of claims 1 to 3, characterized in that: The method further comprises: The facial tags in the face recognition result are parsed, and the facial tags are cached or counted according to the access rights corresponding to the facial tags.

5. An artificial intelligence face recognition system, characterized in that: include: An acquisition module, used to acquire first facial image data obtained by photographing a target face at a first angle and facial adjustment features matching the target face; Among them, the facial adjustment features are additional facial features generated after the face is beautified, covered or worn; a calling module, configured to adjust the first facial feature identified in the first facial image data based on the facial adjustment feature, generate second facial image data, and call a learning coefficient matching the third facial image data obtained by shooting at a second angle and the second facial image data, wherein the learning coefficient is used to characterize the learning intensity of the model training; A training module, used for training an initial facial feature recognition model based on the first facial image data, the second facial image data, the third facial image data and the learning coefficient to obtain a facial feature recognition model that has completed model training; A recognition module, used to perform recognition processing on the collected face image based on the facial feature recognition model to obtain a face recognition result; The step of acquiring first facial image data obtained by photographing a target face at a first angle and facial adjustment features matching the target face comprises: Displaying a first angle, and acquiring first facial image data obtained by photographing a target face at the first angle, wherein the first angle includes at least four directions of shooting angles; Displaying facial feature adjustment items, and determining facial adjustment features through adjustment items selected by the target user from the facial feature adjustment items, wherein the facial feature adjustment items include content items that need to be adjusted for different facial features; and retrieving a learning coefficient that matches the third facial image data obtained by shooting at a second angle and the second facial image data includes: Acquire third facial image data captured at a second angle, where the second angle is obtained by adjusting a preset angle value based on the first angle; adjusting a learning coefficient that matches the second facial image data and the third facial image data based on a preset learning feature adjustment mapping relationship, wherein the preset learning feature adjustment mapping relationship stores learning coefficients corresponding to different facial feature comparison results; The learning coefficients for adjusting the mapping relationship based on the preset learning features to match the second facial image data and the third facial image data include: extracting facial features from the second facial image data and the third facial image data, and comparing the facial features according to pixel features and color features to obtain a facial feature comparison result; Based on the facial feature comparison result, a matching learning coefficient is queried from the preset learning feature adjustment mapping relationship.

6. A storage medium storing at least one executable instruction, wherein the executable instruction enables a processor to perform operations corresponding to the artificial intelligence face recognition method as described in any one of claims 1 to 3.

7. A terminal, comprising: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the artificial intelligence face recognition method according to any one of claims 1-3.

Images