Data audit method, device, computer equipment and storage medium for real-person authentication

Abstract

The embodiment of the present invention discloses a data audit method, device, computer equipment and storage medium for real-person authentication. The method includes: if a video to be detected uploaded by a real-person authentication system is detected, the video to be detected is input into a preset audit model; the key frame images in the video to be detected are extracted through the preset audit model, and the images with identity card information and watermarks in the key frame images are identified, and the confidence of the target image features is calculated through the preset audit model as the target image features, and it is determined whether the confidence is greater than a first preset threshold; if the confidence is greater than the first preset threshold, the video to be detected is confirmed as a non-compliant video. The present invention can not only automatically audit videos during the real-person authentication process, but also improve the ability of automatic auditing and improve efficiency.

Description

Technical Field

[0001] The present invention relates to the field of information processing technology, and in particular to a data audit method, device, computer equipment and storage medium for real-person authentication. Background Art

[0002] The real person authentication service is to strengthen the registration and management of the real identity information of telephone users, requiring telecommunications operators to truthfully register the real identity information of users when handling network access procedures for telephone users. The real person registration process includes steps such as checking the consistency of the person's ID card, checking the authenticity of the ID card, and taking photos on the spot for record.

[0003] In the online video real-person authentication service, users are required to use their mobile phones to shoot video files in real time on the spot to complete the requirements of on-site photo retention and identity verification. However, it is currently found that some users in the online video real-person authentication service use some technical means to synthesize photos into dynamic videos to forge the user's on-site photos. This behavior has the risk of bypassing the user's on-site photo taking, which will cause personal user information to be stolen by illegal elements for real-person authentication services. Summary of the invention

[0004] The embodiments of the present invention provide a data audit method, device, computer equipment and storage medium for real-person authentication, which improve the audit accuracy in the real-person authentication process.

[0005] In a first aspect, an embodiment of the present invention provides a data audit method for real-person authentication, the method comprising:

[0006] If a video to be detected uploaded by the real person authentication system is detected, the video to be detected is input into a preset audit model;

[0007] Extract key frame images in the video to be detected through the preset audit model, identify images with ID card information and watermarks in the key frame images, and use them as target image features;

[0008] Calculating the confidence of the target image feature by using the preset audit model, and determining whether the confidence is greater than a first preset threshold;

[0009] If the confidence level is greater than the first preset threshold, the video to be detected is confirmed as a non-compliant video.

[0010] In a second aspect, an embodiment of the present invention further provides a data auditing device for real-person authentication, the device comprising:

[0011] A first detection unit is used to input the video to be detected into a preset audit model if the video to be detected uploaded by the real person authentication system is detected;

[0012] A first recognition unit is used to extract key frame images in the video to be detected through the preset audit model, and identify the images with identity card information and watermarks in the key frame images, and use them as target image features;

[0013] A first judgment unit, used to calculate the confidence of the target image feature through the preset audit model, and judge whether the confidence is greater than a first preset threshold;

[0014] The first confirmation unit is used to confirm the video to be detected as a non-compliant video if the confidence level is greater than the first preset threshold.

[0015] In a third aspect, an embodiment of the present invention further provides a computer device, which includes a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above method when executing the computer program.

[0016] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the storage medium stores a computer program, and the computer program can implement the above method when executed by a processor.

[0017] The embodiment of the present invention provides a data audit method, device, computer equipment and storage medium for real-person authentication. The method includes: if a video to be detected uploaded by a real-person authentication system is detected, the video to be detected is input into a preset audit model; the key frame picture in the video to be detected is extracted by the preset audit model, and the picture with identity card information and watermark in the key frame picture is identified, and it is used as a target image feature; the confidence of the target image feature is calculated by the preset audit model, and it is judged whether the confidence is greater than a first preset threshold; if the confidence is greater than the first preset threshold, the video to be detected is confirmed as a non-compliant video. The embodiment of the present invention can identify the picture with identity card information and watermark in the video to be detected by the preset audit model when detecting the video to be detected, and use it as a target image feature to calculate the confidence, and judge whether there is non-compliant information or whether there is a watermark in the identity card information according to the size of the confidence and the first preset threshold. When the confidence is greater than the first preset threshold, the video to be detected is judged as a non-compliant video, thereby improving the accuracy of automatic auditing in real-person authentication. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use 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 ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0019] Figure 1 It is a flowchart of a data audit method for real-person authentication provided by an embodiment of the present invention;

[0020] Figure 2 is a schematic block diagram of a data audit device for real-person authentication provided by an embodiment of the present invention;

[0021] Figure 3 It is a schematic block diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, operations, elements, components and / or groups thereof.

[0024] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, the singular forms of "a", "an" and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should also be further understood that the term "and / or" used in the specification of the present invention and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0025] See also Figure 1 , Figure 1 1 is a flow chart of a data audit method for real-person authentication provided by an embodiment of the present invention. The data audit method for real-person authentication provided by an embodiment of the present invention is applicable to any computer device. Figure 1 As shown, the method includes steps S110 to S140.

[0026] S110: If the video to be detected uploaded by the real person authentication system is detected, the video to be detected is input into a preset audit model.

[0027] In the embodiment of the present invention, real person authentication means that when user authorization is required or the user's personal information is verified, the user needs to hold his / her ID card to record a video and upload the recorded video, which is reviewed by a background program or manually. The uploaded video is the video to be detected. After the video to be detected is detected, the video to be detected is input into a preset audit model, and the compliance of the video to be detected is judged by the preset audit model. The main contents include but are not limited to whether the video is shot by the user, whether the video content is clear, and whether it is a synthetic video.

[0028] In certain embodiments, such as the present embodiment, the data audit method for real-person authentication further includes: obtaining a preset training set, wherein the preset training set includes a plurality of labeled images; training the preset training set using a neural network model and labeled images in the preset training set to obtain the preset audit model.

[0029] Among them, before the step of obtaining the preset training set, the following steps are also included: obtaining a preset training video and extracting all optical flow video frames in the preset training video; taking the optical flow video frame with the least number of optical flow movements among all the optical flow video frames as a key frame; marking the area with the identity card information and the watermark in the key frame to obtain the marked picture, and taking all the marked pictures as the preset training set.

[0030] In the embodiment of the present invention, when recording, the user generally holds the ID card and keeps it still for a period of time. That is, under normal circumstances, except for the necessary physiological movements of the human body, such as blinking, the video recorded by the user is static. In particular, for the ID card information, except for shaking, its position will not change significantly. At the same time, if it is a real-time video, there will be no other watermarks except for the environmental background. Therefore, the video uploaded by the user in the past can be used as a preset training video, and the optical flow video frame in the preset training video can be extracted, wherein the optical flow can be understood as the picture of the action change in the video. Among all the optical flow video frames, the optical flow video frame with the least optical flow movement times is used as the key frame, and then the area with the ID card information and watermark in the key frame is marked. In order to improve the accuracy, the user can manually mark it, so as to obtain the preset training set. Then, the area with the ID card information and watermark in the preset training set is trained by the neural network model to obtain the initial training model.

[0031] In certain embodiments, such as the present embodiment, the step of training the preset training set through the neural network model and the annotated pictures in the preset training set to obtain the preset audit model also includes the following steps: obtaining a preset sample set to be identified, training the initial training model through the YOLOv5 neural network model and the preset sample set to be identified to obtain a positive sample set and a negative sample set; training the initial training model through the positive sample set and the negative sample set to obtain the preset audit model; the step of training the initial training model through the positive sample set and the negative sample set to obtain the preset audit model also includes the following steps: training the initial training model through the negative sample set to obtain an intermediate training model; randomly generating the watermark on the pictures in the positive sample set to obtain a target positive sample set; training the intermediate training model through the target positive sample set to obtain the preset audit model.

[0032] In an embodiment of the present invention, after obtaining the initial training model, it is necessary to improve the accuracy of the initial training model. A preset sample set to be identified can be placed in the initial training model so that the initial training model can be trained through the YOLOv5 neural network model to obtain corresponding training results. The training results can be divided into two categories. One is that the initial training model accurately determines the non-compliant videos in the samples to be identified, and the videos are taken as positive sample sets. The other is that the initial training model fails to accurately identify the non-compliant videos in the samples to be identified, and the videos are taken as negative sample sets. The negative sample set is placed in the initial training model again for training until the initial training model can accurately identify all non-compliant videos, and the initial training model is set as an intermediate training model. For the positive sample set, some non-compliant watermarks can be generated or some marks can be marked on the ID card information, and then the intermediate training model can be used to identify and obtain the preset audit model. If the negative sample set can also be obtained after the intermediate training model is trained with the positive sample set, the training is continued with reference to the method described above until no negative sample set is obtained.

[0033] S120, extracting key frame images in the video to be detected through the preset audit model, identifying images with identity card information and watermarks in the key frame images, and using them as target image features.

[0034] In certain embodiments, such as the present embodiment, the step S120 may include the following steps: extracting the key frame image in the video to be detected through the preset audit model and according to preset extraction rules; identifying the area in the key frame image containing the ID card information and the watermark and using it as the target image feature.

[0035] In the embodiment of the present invention, in the video synthesized by photos, generally, there is no way to accurately overlap the position of the ID card, that is, different photos have different positions of the ID card. Similarly, the information carried by the area where the ID card is located will also be different, so the area where the ID card information is located can be extracted as the target image feature. In addition, generally, the video is synthesized by photos, and in most cases, it is necessary to use software for synthesis. When the software is used for synthesis, it will carry the watermark of the software. The position of the watermark can be located in the area where the ID card is located, or it can be any area in the video, so the area in the key frame picture that is obviously inconsistent with the background can be identified as the target image feature. It should be noted that when extracting the key frame picture of the video to be detected, several frames can be randomly selected as the key frame pictures, or several frames in the video can be extracted according to the rules preset by the user.

[0036] S130, calculating the confidence of the target image feature through the preset audit model, and determining whether the confidence is greater than a first preset threshold.

[0037] In an embodiment of the present invention, the preset audit model can calculate the confidence of the target image feature, and judge whether the video to be detected is compliant by judging the size of the confidence and the first preset threshold. Among them, the method of calculating the confidence is well known to those skilled in the art and will not be described here.

[0038] In some embodiments, such as the present embodiment, the step S130 may include the following steps: if the confidence level is less than the second preset threshold, the video to be detected is confirmed as a compliant video; if the confidence level is greater than the second preset threshold and less than the first preset threshold, it is reported to the user for manual review.

[0039] In an embodiment of the present invention, the first preset threshold may be 0.95, and the second preset threshold may be 0.1. When the confidence level is less than 0.1, the application may pass the review directly. When the confidence level is between 0.95 and 0.1, the application may be reported to the user for manual review.

[0040] S140: If the confidence level is greater than the first preset threshold, the video to be detected is confirmed as a non-compliant video.

[0041] In the embodiment of the present invention, when the confidence level is greater than a first preset threshold, for example, greater than 0.95, the video is determined to be a non-compliant video and does not pass the review.

[0042] Figure 2 FIG. 1 is a schematic block diagram of a data audit device 100 for real-person authentication provided by an embodiment of the present invention. Figure 2As shown, corresponding to the above real person authentication data audit method, the present invention also provides a real person authentication data audit device 100. The real person authentication data audit device 100 includes a unit for executing the above real person authentication data audit method. Figure 2 The data auditing device 100 for real person authentication includes a first detection unit 110 , a first identification unit 120 , a first judgment unit 130 and a first confirmation unit 140 .

[0043] Among them, the first detection unit 110 is used to input the video to be detected uploaded by the real-person authentication system into a preset audit model; the first recognition unit 120 is used to extract the key frame pictures in the video to be detected through the preset audit model, and identify the pictures with identity card information and watermarks in the key frame pictures, and use them as target image features; the first judgment unit 130 calculates the confidence of the target image features through the preset audit model, and judges whether the confidence is greater than a first preset threshold; the first confirmation unit 140 is used to confirm the video to be detected as a non-compliant video if the confidence is greater than the first preset threshold.

[0044] In some embodiments, such as the present embodiment, the data auditing device for real-person authentication further includes a first acquisition unit and a first training unit.

[0045] Among them, the first acquisition unit is used to obtain a preset training set, wherein the preset training set includes multiple labeled pictures; the first training unit is used to train the preset training set through a neural network model and the labeled pictures in the preset training set to obtain the preset audit model.

[0046] In some embodiments, such as the present embodiment, the data auditing device for real-person authentication further includes a second acquisition unit, a second confirmation unit and a first marking unit.

[0047] Among them, the second acquisition unit is used to obtain a preset training video and extract all optical flow video frames in the preset training video; the second confirmation unit is used to take the optical flow video frame with the least optical flow movement times among all the optical flow video frames as a key frame; the first labeling unit is used to label the area with the ID card information and the watermark in the key frame to obtain the labeled picture, and take all the labeled pictures as the preset training set.

[0048] In some embodiments, such as this embodiment, the first labeling unit also includes a second recognition unit and a second training unit.

[0049] Among them, the second recognition unit is used to obtain a preset sample set to be identified, and train the initial training model through the YOLOv5 neural network model and the preset sample set to be identified to obtain a positive sample set and a negative sample set; the second training unit is used to train the initial training model through the positive sample set and the negative sample set to obtain the preset audit model.

[0050] In some embodiments, such as this embodiment, the second training unit further includes a third training unit, a third acquisition unit and a fourth training unit.

[0051] Among them, the third training unit is used to train the initial training model through the negative sample set to obtain the intermediate training model; the third acquisition unit is used to randomly generate the watermark on the picture in the positive sample set to obtain the target positive sample set; the fourth training unit is used to train the intermediate training model through the target positive sample set to obtain the preset audit model.

[0052] In some embodiments, such as this embodiment, the first recognition unit 120 includes a first extraction unit and a third recognition unit.

[0053] Among them, the first extraction unit is used to extract the key frame picture in the video to be detected through the preset audit model and according to the preset extraction rules; the third recognition unit is used to identify the area with the ID card information and the watermark in the key frame picture and use it as the target image feature.

[0054] In some embodiments, such as the present embodiment, the data auditing device for real-person authentication further includes a second judgment unit and a third confirmation unit.

[0055] Among them, the second judgment unit is used to confirm the video to be detected as a compliant video if the confidence is less than the second preset threshold; the third confirmation unit is used to report to the user for manual review if the confidence is greater than the second preset threshold and less than the first preset threshold.

[0056] It should be noted that technical personnel in the relevant field can clearly understand that the specific implementation process of the data audit device and each unit of the above-mentioned real-person authentication can refer to the corresponding description in the aforementioned method embodiment, and for the convenience and conciseness of the description, it will not be repeated here.

[0057] The data auditing device for real person authentication can be implemented in the form of a computer program. Figure 3 Runs on the computer device shown.

[0058] See also Figure 3 , Figure 3is a schematic block diagram of a computer device provided in an embodiment of the present application. Figure 3 The computer device 500 includes a processor 502 , a memory and an interface 507 connected via a system bus 501 , wherein the memory may include a non-volatile storage medium 503 and an internal memory 504 .

[0059] The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, the processor 502 can execute a data audit method for real person authentication.

[0060] The processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500 .

[0061] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute any embodiment of the data audit method for real-person authentication.

[0062] The interface 505 is used to communicate with other devices. Those skilled in the art will appreciate that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present application, and does not constitute a limitation on the computer device 500 to which the solution of the present application is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0063] The processor 502 is used to run the computer program 5032 stored in the memory to implement the following steps:

[0064] If a video to be detected uploaded by the real person authentication system is detected, the video to be detected is input into a preset audit model;

[0065] Extract key frame images in the video to be detected through the preset audit model, identify images with ID card information and watermarks in the key frame images, and use them as target image features;

[0066] Calculating the confidence of the target image feature by using the preset audit model, and determining whether the confidence is greater than a first preset threshold;

[0067] If the confidence level is greater than the first preset threshold, the video to be detected is confirmed as a non-compliant video.

[0068] In one embodiment, the processor 502 further implements the following steps:

[0069] Obtain a preset sample set to be identified, and train the initial training model using a YOLOv5 neural network model and the preset sample set to be identified to obtain a positive sample set and a negative sample set;

[0070] The initial training model is trained using the positive sample set and the negative sample set to obtain the preset audit model.

[0071] In one embodiment, when the processor 502 implements the step of obtaining the preset training set, it specifically implements the following steps:

[0072] Obtain a preset training video, and extract all optical flow video frames in the preset training video;

[0073] The optical flow video frame with the least number of optical flow movements among all the optical flow video frames is used as a key frame;

[0074] The area with the identity card information and the watermark in the key frame is marked to obtain the marked picture, and all the marked pictures are used as the preset training set.

[0075] In one embodiment, when the processor 502 implements the step of training the preset training set by using the neural network model and the annotated pictures in the preset training set to obtain the preset audit model, the processor 502 specifically implements the following steps:

[0076] Obtain a preset sample set to be identified, and train the initial training model using a YOLOv5 neural network model and the preset sample set to be identified to obtain a positive sample set and a negative sample set;

[0077] The initial training model is trained using the positive sample set and the negative sample set to obtain the preset audit model.

[0078] In one embodiment, when the processor 502 implements the step of training the initial training model by using the positive sample set and the negative sample set to obtain the preset audit model, the processor 502 specifically implements the following steps:

[0079] Training the initial training model using the negative sample set to obtain an intermediate training model;

[0080] Randomly generate the watermark on the pictures in the positive sample set to obtain a target positive sample set;

[0081] The intermediate training model is trained using the target positive sample set to obtain the preset audit model.

[0082] In one embodiment, when the processor 502 implements the step of extracting the key frame pictures in the video to be detected by using the preset audit model, identifying the pictures with the ID card information and the watermark in the key frame pictures, and using them as the target image features, the processor 502 specifically implements the following steps:

[0083] Extracting key frame images in the video to be detected by using the preset audit model and according to preset extraction rules;

[0084] The area with the ID card information and the watermark in the key frame image is identified and used as the target image feature.

[0085] In one embodiment, after the processor 502 implements the step of calculating the confidence of the target image feature by using the preset audit model and determining whether the confidence is greater than a first preset threshold, the processor 502 further includes the following steps:

[0086] If the confidence level is less than the second preset threshold, the video to be detected is confirmed as a compliant video;

[0087] If the confidence level is greater than the second preset threshold and less than the first preset threshold, the user is reported for manual review.

[0088] It should be understood that in the embodiment of the present application, the processor 502 may be a central processing unit (CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (FSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0089] It is understood by those skilled in the art that all or part of the processes in the method for implementing the above embodiment can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the embodiment of the above method.

[0090] Therefore, the present invention also provides a storage medium. The storage medium may be a computer-readable storage medium. The storage medium stores a computer program. When the computer program is executed by a processor, any embodiment of the data audit method for real person authentication is implemented.

[0091] The storage medium may be a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disk, or any other computer-readable storage medium that can store program codes.

[0092] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0093] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of each unit is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0094] The steps in the method of the embodiment of the present invention can be adjusted in order, combined and deleted according to actual needs. The units in the device of the embodiment of the present invention can be combined, divided and deleted according to actual needs. In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0095] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for a computer device to execute all or part of the steps of the method described in each embodiment of the present invention.

[0096] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0097] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0098] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A data audit method for real person authentication, characterized in that: The method comprises: If a video to be detected uploaded by the real person authentication system is detected, the video to be detected is input into a preset audit model; Extract key frame images in the video to be detected through the preset audit model, identify images with ID card information and watermarks in the key frame images, and use them as target image features; Calculating the confidence of the target image feature by using the preset audit model, and determining whether the confidence is greater than a first preset threshold; If the confidence level is greater than the first preset threshold, the video to be detected is confirmed as a non-compliant video; Obtaining a preset training set, wherein the preset training set includes a plurality of annotated pictures; The preset training set is trained using a neural network model and labeled images in the preset training set to obtain the preset audit model.

2. The data audit method for real person authentication as claimed in claim 1, characterized in that: Before the step of obtaining the preset training set, the method further includes: Obtain a preset training video, and extract all optical flow video frames in the preset training video; The optical flow video frame with the least number of optical flow movements among all the optical flow video frames is used as a key frame; The area with the identity card information and the watermark in the key frame is marked to obtain the marked picture, and all the marked pictures are used as the preset training set.

3. The data audit method for real person authentication as claimed in claim 1, characterized in that: The step of training the preset training set by using the neural network model and the annotated pictures in the preset training set to obtain the preset audit model includes: Obtain a preset sample set to be identified, and train an initial training model using a YOLOv5 neural network model and the preset sample set to be identified to obtain a positive sample set and a negative sample set; The initial training model is trained using the positive sample set and the negative sample set to obtain the preset audit model.

4. The data audit method for real person authentication as claimed in claim 3, characterized in that: The step of training the initial training model by using the positive sample set and the negative sample set to obtain the preset audit model comprises: Training the initial training model using the negative sample set to obtain an intermediate training model; Randomly generate the watermark on the pictures in the positive sample set to obtain a target positive sample set; The intermediate training model is trained using the target positive sample set to obtain the preset audit model.

5. The data audit method for real person authentication as claimed in claim 2, characterized in that: The step of extracting key frame images in the video to be detected by using the preset audit model, identifying images with identity card information and watermarks in the key frame images, and using them as target image features includes: Extracting key frame images in the video to be detected by using the preset audit model and according to preset extraction rules; The area with the ID card information and the watermark in the key frame image is identified and used as the target image feature.

6. The data audit method for real person authentication as claimed in claim 1, characterized in that: After the step of calculating the confidence of the target image feature by the preset audit model and determining whether the confidence is greater than a first preset threshold, the method further includes: If the confidence level is less than a second preset threshold, the video to be detected is confirmed as a compliant video; If the confidence level is greater than the second preset threshold and less than the first preset threshold, the user is reported for manual review.

7. A data auditing device for real person authentication, characterized in that: The device comprises: A first detection unit is used to input the video to be detected into a preset audit model if the video to be detected uploaded by the real person authentication system is detected; A first recognition unit is used to extract key frame images in the video to be detected through the preset audit model, and identify the images with identity card information and watermarks in the key frame images, and use them as target image features; A first judgment unit, used to calculate the confidence of the target image feature through the preset audit model, and judge whether the confidence is greater than a first preset threshold; A first confirmation unit, configured to confirm the video to be detected as a non-compliant video if the confidence level is greater than the first preset threshold; A first acquisition unit, configured to acquire a preset training set, wherein the preset training set includes a plurality of annotated pictures; The first training unit is used to train the preset training set through a neural network model and the labeled pictures in the preset training set to obtain the preset audit model.

8. A computer device, characterized in that: The computer device comprises a memory and a processor connected to the memory; the memory is used to store a computer program; the processor is used to run the computer program stored in the memory to execute the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 can be implemented on a computer device.

Images