A face image quality perception threshold setting method and device and a storage medium

Abstract

A face image quality perception threshold setting method and device and a storage medium, the method comprising the steps of: constructing a face image sample set; constructing a model according to the face image sample set; performing face detection on an input image and obtaining a face image; performing face image quality evaluation on the face image; and setting a recognition pass threshold according to the face image quality evaluation. The face image quality perception threshold setting method, device and storage medium provided by the application can be used in face recognition applications in non-constrained and non-cooperative scenarios, control the false recognition rate of the to-be-recognized image and reduce the rejection rate while maintaining the face recognition accuracy, and improve the recognition speed, thereby improving the efficiency of the face recognition system.

Description

Technical Field

[0001] This invention belongs to the field of facial image recognition technology, specifically relating to a method, device, and storage medium for setting a facial image quality perception threshold. Background Technology

[0002] Currently, the popular solutions to the problem of high rejection rates in face recognition due to the large number of low-quality face images in real-world scenarios can be divided into two main types: (1) Establishing corresponding clustering features and scene classification models based on the scene to which the face image belongs, and predicting the scene category of the face image to be recognized, thereby selecting the face clustering features to be matched for face recognition based on the scene category; (2) Dynamically adjusting the recognition threshold based on the image quality difference between the face image to be recognized and the face images registered in the database, while continuously updating the face database with the input image to be recognized, thereby realizing the dynamic adjustment of the robustness and adaptability of the face recognition system to different environments and image quality.

[0003] In method (1), specific scene clustering features are selected for matching through quality assessment of the face image to be identified. This has a high recognition accuracy and low rejection rate for existing scenes. However, in order to achieve higher face recognition accuracy, a more complex clustered face feature library needs to be established. Furthermore, for new quality scenes, its generalization ability is limited, and new scene clustering features need to be added to improve recognition accuracy. Method (2) sets a series of recognition thresholds from low to high to perform step-by-step feature matching on the face to be identified, thereby reducing the rejection rate of the face image to be identified. However, there are multiple repeated feature and image quality difference calculation processes in the step-by-step matching process, which has an adverse effect on the real-time performance of the face recognition system and reduces the user experience. Summary of the Invention

[0004] In view of the above problems, the present invention provides a method, apparatus and storage medium for setting a face image quality perception threshold to overcome the above problems or at least partially solve the above problems.

[0005] To address the aforementioned technical problems, this invention provides a method for facial image quality recognition through threshold setting, the method comprising the following steps:

[0006] Construct a sample set of human face images;

[0007] A model is constructed based on the aforementioned set of face image samples;

[0008] Perform face detection on the input image and obtain face images;

[0009] The facial image quality is evaluated.

[0010] The recognition threshold is set based on the facial image quality evaluation.

[0011] Preferably, constructing the face image sample set includes the following steps:

[0012] Collect face images of different quality levels and scene categories;

[0013] The quality level of the face image is labeled;

[0014] The face images are divided into sample sets according to the quality scene category and the quality level label;

[0015] The optimal recognition threshold is set for the face image based on the sample set division;

[0016] Face training samples are constructed using the face image, the quality scene category, the quality level label, and the optimal recognition through a threshold.

[0017] All the face training samples are summarized to obtain the face image sample set.

[0018] Preferably, the step of constructing a model based on the face image sample set includes the following steps:

[0019] Obtain the face image sample set, the natural image quality assessment model, the face images, and the face recognition dataset;

[0020] A face image quality assessment model is constructed based on the face image sample set and the natural image quality assessment model;

[0021] The face image is input into the face image quality assessment model to obtain the face image quality scene category and face image quality assessment grade;

[0022] A threshold mapping model is constructed based on the aforementioned face image sample set;

[0023] The face image quality scene category and the face image quality evaluation level are input into the threshold mapping model to obtain the recognition threshold;

[0024] A face recognition model is constructed based on the face recognition dataset.

[0025] Preferably, the step of constructing a face image quality assessment model based on the face image sample set and the natural image quality assessment model includes the following steps:

[0026] A prior model for face image quality assessment is established based on the natural image quality assessment model described above.

[0027] The prior model for evaluating the quality of facial images is adjusted based on the set of facial image samples.

[0028] Preferably, the step of constructing a threshold mapping model based on the face image sample set includes the following steps:

[0029] Build the basic model;

[0030] Initialize the parameters of the basic model;

[0031] The base model is trained and optimized based on the face image sample set to obtain a threshold mapping model.

[0032] Preferably, the step of establishing a prior model for face image quality assessment based on the natural image quality assessment model includes the following steps:

[0033] Obtain the face image;

[0034] Preliminary feature extraction for facial quality assessment is performed on the facial image;

[0035] An adaptive max pooling operation is performed on the preliminary feature extraction for the face quality assessment.

[0036] The quality scene category and quality assessment grade of the face image are generated based on the adaptive max pooling operation.

[0037] Preferably, adjusting the prior model for facial image quality evaluation based on the facial image sample set includes the following steps:

[0038] The parameters of the prior model for evaluating the quality of the face image are initialized;

[0039] Obtain the face image sample set;

[0040] The face image quality assessment prior model is trained and optimized using the face image sample set.

[0041] This application also provides a face image quality recognition device based on a threshold setting, the device comprising:

[0042] The face image sample set construction module is used to construct face image sample sets;

[0043] The model building module is used to build a model based on the face image sample set;

[0044] The face image acquisition module is used to detect faces in the input image and obtain face images;

[0045] A face image quality evaluation module is used to evaluate the face image quality.

[0046] The recognition threshold setting module is used to set the recognition threshold based on the quality evaluation of the face image.

[0047] This application also provides an electronic device, the electronic device comprising:

[0048] At least one processor; and,

[0049] A memory communicatively connected to the at least one processor; wherein,

[0050] The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform any of the aforementioned face image quality perception threshold setting methods.

[0051] This application also provides a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute any of the aforementioned face image quality perception threshold setting methods.

[0052] One or more technical solutions in the embodiments of the present invention have at least the following technical effects or advantages: The face image quality perception threshold setting method, device and storage medium provided in this application can be used in face recognition applications in unconstrained and non-cooperative scenarios to control the false recognition rate and reduce the rejection rate of the image to be recognized while maintaining the accuracy of face recognition, and at the same time improve the recognition speed, thereby improving the efficiency of the face recognition system. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a flowchart illustrating a method for setting a human face image quality perception threshold provided by the present invention.

[0055] Figure 2 This is a schematic diagram of the structure of a face image quality perception threshold setting device provided by the present invention;

[0056] Figure 3 This is a schematic diagram of the structure of an electronic device provided by the present invention;

[0057] Figure 4 This is a schematic diagram of the structure of a non-transitory computer-readable storage medium provided by the present invention. Detailed Implementation

[0058] The present invention will be described in detail below with reference to specific embodiments and examples, thereby making the advantages and various effects of the present invention more clearly apparent. Those skilled in the art should understand that these specific embodiments and examples are for illustrative purposes only and are not intended to limit the present invention.

[0059] Throughout this specification, unless otherwise specified, the terminology used herein should be understood as having the meaning commonly used in the art. Therefore, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. In the event of any conflict, this specification shall prevail.

[0060] Unless otherwise specified, all raw materials, reagents, instruments and equipment used in this invention can be purchased from the market or prepared by existing methods.

[0061] like Figure 1 In this embodiment, the present invention provides a method for face image quality recognition by setting a threshold, the method comprising the steps of:

[0062] S1: Construct a sample set of face images;

[0063] In this embodiment of the application, the construction of the face image sample set includes the following steps:

[0064] Collect face images of different quality levels and scene categories;

[0065] The quality level of the face image is labeled;

[0066] The face images are divided into sample sets according to the quality scene category and the quality level label;

[0067] The optimal recognition threshold is set for the face image based on the sample set division;

[0068] Face training samples are constructed using the face image, the quality scene category, the quality level label, and the optimal recognition through a threshold.

[0069] All the face training samples are summarized to obtain the face image sample set.

[0070] In this embodiment, the face image sample set is created through the following steps: First, face images under different quality scene categories are collected, such as face images A, B, and C under bright light scene category, normal light scene category, and dark light scene category. Then, the quality levels of face images A, B, and C are labeled (based on the face image quality standards ISO / IEC 19794-5 and ISO / IEC 29794-5 published by the International Organization for Standardization), which are H (high quality), H (high quality), and L (low quality), respectively. Next, the face images are divided into sample sets according to the quality scene category and quality level label. For example, since the quality scene categories of face images A, B, and C are bright light scene category, normal light scene category, and dark light scene category, respectively, and the quality level labels are H, H, and L, respectively, A and B can be divided into one sample set, denoted as (A, B), and C can be divided into another sample set, denoted as (C). The division criteria can be set as needed. Then, based on the sample set division, an optimal recognition threshold is set for the face images. For example, the optimal recognition threshold can be set to D (the optimal recognition threshold is set for the face images in the sample set based on the ROC curve of each sample set on the face recognition model). Finally, face training samples are constructed based on the face images, the quality scene category, the quality level label, and the optimal recognition threshold. All the face training samples are then summarized to obtain the face image sample set.

[0071] S2: Construct a model based on the face image sample set;

[0072] In this embodiment of the application, the step of constructing a model based on the face image sample set includes the following steps:

[0073] Obtain the face image sample set, the natural image quality assessment model, the face images, and the face recognition dataset;

[0074] A face image quality assessment model is constructed based on the face image sample set and the natural image quality assessment model;

[0075] The face image is input into the face image quality assessment model to obtain the face image quality scene category and face image quality assessment grade;

[0076] A threshold mapping model is constructed based on the aforementioned face image sample set;

[0077] The face image quality scene category and the face image quality evaluation level are input into the threshold mapping model to obtain the recognition threshold;

[0078] A face recognition model is constructed based on the face recognition dataset.

[0079] In this embodiment, when constructing a model based on the face image sample set, it specifically refers to constructing a face image quality assessment model, a threshold mapping model, and a face recognition model. The face image quality assessment model is constructed based on the face image sample set and a natural image quality assessment model. Inputting a face image into the face image quality assessment model yields the face image quality scene category and the face image quality assessment grade. The threshold mapping model is constructed using the face image sample set. Inputting the face image quality scene category and the face image quality assessment grade into the threshold mapping model yields the recognition pass threshold. The face recognition model is constructed using a face recognition dataset (the face recognition model is obtained by training the high-performance face recognition model ArcFace based on the CelebA face recognition dataset).

[0080] In this embodiment of the application, the step of constructing a face image quality assessment model based on the face image sample set and the natural image quality assessment model includes the following steps:

[0081] A prior model for face image quality assessment is established based on the natural image quality assessment model described above.

[0082] The prior model for evaluating the quality of facial images is adjusted based on the set of facial image samples.

[0083] In this embodiment, when constructing a face image quality assessment model based on a face image sample set and a natural image quality assessment model, a prior model for face image quality assessment is first established based on the natural image quality assessment model, and then the prior model is adjusted based on the face image sample set. Specific steps are further elaborated in subsequent sections.

[0084] In this embodiment of the application, the step of constructing a threshold mapping model based on the face image sample set includes the following steps:

[0085] Build the basic model;

[0086] Initialize the parameters of the basic model;

[0087] The base model is trained and optimized based on the face image sample set to obtain a threshold mapping model.

[0088] In this embodiment of the application, when constructing a threshold mapping model based on the face image sample set, a basic model is first constructed using two fully connected layers and one softmax classifier, and the parameters of the basic model are initialized using the Kaiming initialization method; then the basic model is trained and optimized using the face image sample set, and the threshold mapping model can be obtained.

[0089] In this embodiment of the application, the step of establishing a prior model for face image quality assessment based on the natural image quality assessment model includes the following steps:

[0090] Obtain the face image;

[0091] Preliminary feature extraction for facial quality assessment is performed on the facial image;

[0092] An adaptive max pooling operation is performed on the preliminary feature extraction for the face quality assessment.

[0093] The quality scene category and quality assessment grade of the face image are generated based on the adaptive max pooling operation.

[0094] In this embodiment of the application, when establishing a prior model for face image quality assessment based on the natural image quality assessment model, a nine-layer convolutional neural network is first used to extract preliminary features for face quality assessment of the face image. Then, an adaptive max pooling operation is performed on the output of the nine-layer convolutional neural network (the preliminary feature extraction result for face quality assessment) through an adaptive max pooling layer. The quality scene category of the face image is generated through a fully connected layer and a softmax classifier module. At the same time, the quality assessment grade of the face image is generated through another fully connected layer and a softmax classifier module.

[0095] In this embodiment of the application, adjusting the prior model for facial image quality evaluation based on the facial image sample set includes the following steps:

[0096] The parameters of the prior model for evaluating the quality of the face image are initialized;

[0097] Obtain the face image sample set;

[0098] The face image quality assessment prior model is trained and optimized using the face image sample set.

[0099] In this embodiment, when adjusting the face image quality assessment prior model based on the face image sample set, the model parameters of the natural image quality assessment model DBCNN are first used to initialize the parameters of the face image quality assessment prior model. Then, the face image sample set obtained in step S1 is used to train and optimize the face image quality assessment prior model. Specifically, for an input face image x, the face quality assessment score obtained by inputting x into the face image quality assessment prior model is: Where θ is the parameter of the prior model for face image quality assessment; the L2 norm loss function between the quality score of the input face image predicted by the model and the true quality score is used to optimize the parameters of the prior model for face image quality assessment, and the specific definition of the loss function is: Where y represents the true quality score of the input face image; the prior model for face image quality assessment is optimized using stochastic gradient descent, and the parameters of the prior model for face image quality assessment are updated using the Adam optimizer to obtain the training optimization results.

[0100] S3: Perform face detection on the input image and obtain a face image;

[0101] In this embodiment of the application, when performing face detection on the input image, the Retinaface face detection model is used to detect faces in the input image, mark face boxes, and crop the input image using the face marking boxes to obtain the face image to be identified.

[0102] S4: Evaluate the face image quality of the face image;

[0103] In this embodiment of the application, the input face image is evaluated for face image quality, that is, the quality scene is classified and the quality is graded.

[0104] S5: Set the recognition threshold based on the face image quality evaluation.

[0105] In this embodiment of the application, the recognition of the face recognition model is set by setting a threshold based on the quality scene category and quality rating of the input face image.

[0106] In this embodiment, the DSFD face detection model can also be used to perform face detection on the input image. The DSFD face detection model is a high-performance model for face detection. It achieves feature enhancement through a dual-channel deep residual network, thereby realizing high-precision, multi-scale adaptive face detection.

[0107] In this embodiment of the application, in a multi-class classification task, the softmax classifier obtains the probability of classifying the input x into each category: Where θ is the model parameter, j is the category number, and y is the result category number.

[0108] In this embodiment, the Kaiming initialization method is a model weight initialization method that scales the weight parameters of the network weights initialized according to a standard distribution, with a scaling factor of [missing value]. Where n is the dimension of the input vector, and the bias parameter of the network weights is uniformly initialized to 0.

[0109] In the embodiments of this application, the natural image quality assessment model DBCNN is a dual convolutional network model for image quality assessment. Its model structure consists of a parallel S-CNN model and a VGG-16 model, a pooling layer, and a fully connected layer.

[0110] like Figure 2 This application also provides a face image quality recognition device through threshold setting, the device comprising:

[0111] Face image sample set construction module 10, used to construct face image sample sets;

[0112] Model building module 20 is used to build a model based on the face image sample set;

[0113] The face image acquisition module 30 is used to perform face detection on the input image and obtain a face image;

[0114] The face image quality evaluation module 40 is used to evaluate the face image quality.

[0115] The recognition threshold setting module 50 is used to set the recognition threshold based on the face image quality evaluation.

[0116] The face image quality perception threshold setting device provided in this application can perform the face image quality perception threshold setting method provided in the above steps.

[0117] The following is for reference. Figure 3 The diagram illustrates a structural schematic of an electronic device 100 suitable for implementing embodiments of the present disclosure. The electronic devices in the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0118] like Figure 3 As shown, the electronic device 100 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 101, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 102 or a program loaded from a storage device 108 into a random access memory (RAM) 103. The RAM 103 also stores various programs and data required for the operation of the electronic device 100. The processing unit 101, ROM 102, and RAM 103 are interconnected via a bus 104. An input / output (I / O) interface 105 is also connected to the bus 104.

[0119] Typically, the following devices can be connected to I / O interface 105: input devices 106 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 107 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 108 including, for example, magnetic tapes, hard disks, etc.; and communication devices 109. Communication device 109 allows electronic device 100 to communicate wirelessly or wiredly with other devices to exchange data. Although electronic device 100 with various devices is shown in the figure, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0120] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 109, or installed from storage device 108, or installed from ROM 102. When the computer program is executed by processing device 101, it performs the functions defined in the methods of embodiments of this disclosure.

[0121] The following is for reference. Figure 4 The diagram illustrates a computer-readable storage medium suitable for implementing embodiments of the present disclosure, the computer-readable storage medium storing a computer program that, when executed by a processor, can implement the face image quality perception threshold setting method as described above.

[0122] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0123] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0124] The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: acquire at least two Internet Protocol (IP) addresses; send a node evaluation request including the at least two IP addresses to a node evaluation device, wherein the node evaluation device selects an IP address from the at least two IP addresses and returns it; and receive the IP address returned by the node evaluation device; wherein the acquired IP address indicates an edge node in a content delivery network.

[0125] Alternatively, the aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: receive a node evaluation request including at least two Internet Protocol (IP) addresses; select an IP address from the at least two IP addresses; and return the selected IP address; wherein the received IP address indicates an edge node in the content delivery network.

[0126] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0127] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0128] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The name of a unit does not necessarily limit the unit itself; for example, the first acquisition unit can also be described as "a unit that acquires at least two Internet Protocol addresses".

[0129] This application provides a method, apparatus, and storage medium for setting a face image quality perception threshold, which can be used in face recognition applications in unconstrained and non-cooperative scenarios to control the false recognition rate and reduce the rejection rate of the image to be recognized while maintaining the accuracy of face recognition, and at the same time improve the recognition speed, thereby improving the performance of the face recognition system.

[0130] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of the invention and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of the invention should be included within the protection scope of the invention. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

Claims

1. A method for face image quality recognition by setting a threshold, characterized in that, The method includes the following steps: Construct a sample set of human face images; A model is constructed based on the aforementioned face image sample set; Perform face detection on the input image and obtain face images; The facial image quality is evaluated. The recognition threshold is set based on the facial image quality evaluation. The steps of constructing the model based on the face image sample set include: Obtain the face image sample set, the natural image quality assessment model, the face images, and the face recognition dataset; A face image quality assessment model is constructed based on the face image sample set and the natural image quality assessment model; The face image is input into the face image quality assessment model to obtain the face image quality scene category and face image quality assessment grade; A threshold mapping model is constructed based on the aforementioned face image sample set; The face image quality scene category and the face image quality evaluation level are input into the threshold mapping model to obtain the recognition threshold; Construct a face recognition model based on the aforementioned face recognition dataset; The step of constructing a face image quality assessment model based on the face image sample set and the natural image quality assessment model includes the following steps: A prior model for face image quality assessment is established based on the natural image quality assessment model described above. The prior model for evaluating face image quality is adjusted based on the face image sample set; The step of constructing a threshold mapping model based on the face image sample set includes the following steps: Build the basic model; Initialize the parameters of the basic model; The base model is trained and optimized based on the face image sample set to obtain a threshold mapping model; The steps for establishing a priori model for face image quality assessment based on the natural image quality assessment model include: Obtain the face image; Preliminary feature extraction for facial quality assessment is performed on the facial image; An adaptive max pooling operation is performed on the preliminary feature extraction for the face quality assessment. The quality scene category and quality assessment grade of the face image are generated based on the adaptive max pooling operation; The step of adjusting the prior model for facial image quality assessment based on the facial image sample set includes the following steps: The parameters of the prior model for evaluating the quality of the face image are initialized; Obtain the face image sample set; The face image quality assessment prior model is trained and optimized using the face image sample set.

2. The face image quality recognition method according to claim 1, characterized in that, The steps for constructing the face image sample set are as follows: Collect face images of different quality levels and scene categories; The quality level of the face image is labeled; The face images are divided into sample sets according to the quality scene category and the quality level label; The optimal recognition threshold is set for the face image based on the sample set division; Face training samples are constructed using the face image, the quality scene category, the quality level label, and the optimal recognition through a threshold. All the face training samples are summarized to obtain the face image sample set.

3. A face image quality recognition method using a threshold setting device, applicable to the method described in claim 1 or 2, characterized in that, The device includes: The face image sample set construction module is used to construct face image sample sets; The model building module is used to build a model based on the face image sample set; The face image acquisition module is used to detect faces in the input image and obtain face images; A face image quality evaluation module is used to evaluate the face image quality. The recognition threshold setting module is used to set the recognition threshold based on the quality evaluation of the face image.

4. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the face image quality perception threshold setting method according to claim 1 or 2.

5. A non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the face image quality perception threshold setting method of claim 1 or 2.

Images