A face recognition method, system, electronic device and storage medium

Abstract

The application provides a face recognition method, system, electronic device and storage medium, the face recognition system comprises a characteristic value conversion device, a platform end and a device end, the platform end is connected with the characteristic value conversion device and the device end in communication respectively; the characteristic value conversion device converts at least one face picture obtained from the platform end into corresponding characteristic values and sends the characteristic values to the platform end; the platform end forwards each characteristic value to the device end; the device end converts a target face picture collected into a target characteristic value, and performs face comparison on the target characteristic value and each characteristic value respectively, when the comparison result meets a preset condition, a prompt information is triggered, so as to avoid affecting the system energy of the platform end, and to avoid data leakage, reduce the storage space and network bandwidth occupied by the device end.

Description

Technical Field

[0001] This invention relates to the field of facial recognition technology, and more specifically, to a facial recognition method, system, electronic device, and storage medium. Background Technology

[0002] With the maturity of facial recognition technology, it has been widely used in law enforcement and security monitoring, especially in scenarios such as face scanning and blacklist recognition.

[0003] In existing technologies, a platform can generate feature values ​​for facial images and transmit the facial images and their feature values ​​to a device. The device can then store the facial images and feature values ​​for subsequent facial recognition and comparison using the feature values. When a target face or a face from a blacklist is identified, the device directly displays the stored facial image.

[0004] However, the computation of feature values ​​is relatively large, and performing feature value conversion on the platform can easily affect the system performance of the platform. Moreover, in police scenarios, since blacklisted facial images are sensitive data, directly transmitting them to the device can easily lead to leakage risks. Furthermore, when the amount of facial image data is large, directly transmitting it to the device will not only occupy a lot of storage space on the device, but also consume a lot of network bandwidth, thereby affecting the facial recognition efficiency of the device. Summary of the Invention

[0005] In view of this, the present invention provides a face recognition method, system, electronic device and storage medium, with the aim of avoiding impact on the system energy of the platform, avoiding data leakage, and reducing the storage space and network bandwidth occupied on the device.

[0006] The first aspect of this application provides a face recognition system, which includes a feature value conversion device, a platform terminal, and a device terminal, wherein the platform terminal is communicatively connected to the feature value conversion device and the device terminal respectively;

[0007] The feature value conversion device is used to convert at least one face image obtained from the platform into corresponding feature values ​​and send them to the platform.

[0008] The platform is used to forward each of the feature values ​​to the device.

[0009] The device is used to convert the acquired target face image into target feature values, and compare the target feature values ​​with each of the feature values ​​for face recognition. When the comparison result meets the preset conditions, a prompt message is triggered.

[0010] Optionally, the feature value conversion device, which converts multiple face images obtained from the platform into feature values, is specifically used for:

[0011] The face database task on the platform is periodically queried to obtain at least one face image from the platform based on the face database task; wherein the face image is an encrypted face image;

[0012] The face image is converted into feature values ​​using a preset face recognition algorithm, wherein the face image cannot be reconstructed using the feature values.

[0013] Optionally, the feature value conversion device is further used for:

[0014] Construct a corresponding vector index based on the identifier of the face image;

[0015] Establish a mapping relationship between the vector index and the feature values ​​of the face image, and establish a mapping relationship between the vector index and the identifier of the face image;

[0016] A feature value index file is generated based on the feature values ​​of the face image and their mapping relationship, and a metadata file is generated based on the identifier of the face image and its mapping relationship;

[0017] The feature value index file and the metadata file are sent to the platform.

[0018] Optionally, the device that converts the real-time acquired target face image into target feature values ​​is specifically used for:

[0019] Real-time acquisition of target facial images;

[0020] The target face image is converted into target feature values ​​using a preset face recognition algorithm; wherein the target face image cannot be reconstructed using the target feature values.

[0021] Optionally, the device that performs face comparison between the target feature value and each of the respective feature values ​​is specifically used for:

[0022] Calculate the similarity between the target feature value and each of the feature values;

[0023] Determine whether any of the aforementioned similarities exceeds a preset threshold;

[0024] If it does not exist, generate a comparison result that does not meet the preset conditions;

[0025] If such a similarity exists, the maximum similarity is determined from the similarities greater than the preset threshold, and a comparison result that meets the preset conditions is generated based on the feature value corresponding to the maximum similarity.

[0026] Optionally, the platform terminal is also used for:

[0027] Receive the comparison results sent by the device that meet the preset conditions;

[0028] Obtain the target vector index corresponding to the feature value indicated by the alignment result from the feature value index file;

[0029] Obtain the target identifier that matches the target vector index from the metadata file;

[0030] Obtain a face image that matches the target identifier, wherein the face image that matches the target identifier is a face image whose feature value matches the feature value indicated in the comparison result;

[0031] The face image matching the target identifier is sent to the device.

[0032] Optionally, the platform terminal is also used for:

[0033] Receive an image request sent by the device, wherein the image request includes user information of the user to be verified;

[0034] Based on the user information, determine whether the user to be verified has permission to view facial images;

[0035] If it exists, perform the step of sending the face image that matches the target identifier to the device.

[0036] If not, send a message to the device indicating that the user does not have permission to view the face image.

[0037] A second aspect of this application provides a face recognition method applied to the face recognition system provided in the first aspect of this application, the method comprising:

[0038] The feature value conversion device converts at least one face image obtained from the platform into corresponding feature values ​​and sends them to the platform.

[0039] The platform forwards each of the feature values ​​to the device.

[0040] The device converts the captured target face image into target feature values, and compares the target feature values ​​with each of the feature values. When the comparison result meets the preset conditions, a prompt message is triggered.

[0041] A third aspect of this application provides an electronic device, including: a processor and a memory, the processor and the memory being connected via a bus; wherein, the processor is used to call and execute a program stored in the memory; the memory is used to store the program, the program being used to implement the face recognition method provided in the first aspect of this application.

[0042] A fourth aspect of this application provides a computer-readable storage medium storing computer-executable instructions for performing the face recognition method provided in the first aspect of this application.

[0043] This application provides a face recognition method, system, electronic device, and storage medium. The face recognition system consists of a feature value conversion device, a platform, and a device. The feature value conversion device converts at least one face image acquired from the platform into feature values ​​and sends them to the platform. The platform forwards each feature value to the device. The device converts the acquired target face image into target feature values ​​and compares the target feature values ​​with each of the other feature values. When the comparison result meets preset conditions, a prompt message is triggered. Therefore, the technical solution provided in this application separates the feature value conversion task from the platform to an independent feature value conversion device. This not only decouples the system and reduces the computational pressure on the platform, avoiding impact on the platform's system performance, but also ensures that face images are transmitted only between the platform and the feature value conversion device, thus avoiding the risk of data leakage at the device. Furthermore, the platform provided in this application only sends feature values ​​to the device, significantly reducing data transmission volume. The device provided in this application is responsible for performing the corresponding feature comparisons, eliminating the need to store a large number of face images, thus avoiding the occupation of large amounts of storage space and network bandwidth, thereby improving face recognition efficiency. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0045] Figure 1 This is a schematic diagram of the structure of a face recognition system provided in an embodiment of this application;

[0046] Figure 2 This is a schematic diagram of another face recognition system provided in an embodiment of this application;

[0047] Figure 3 This is a flowchart illustrating a face recognition method provided in an embodiment of the present invention;

[0048] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] In this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0051] To better understand this application, the technical terms used in this application are explained below:

[0052] Face Embedding: Converts a face image into a numerical vector representation for face recognition and comparison.

[0053] Feature value conversion device (Wanted Database Builder, WDB): A stand-alone facial feature value conversion device responsible for converting facial images into feature values, and does not directly participate in the face recognition comparison on the device side.

[0054] Vector Database: A database that stores feature values, including feature value index files (.bin) and metadata files (.metainfo), used for fast similarity matching.

[0055] Wanted Face Database: A database that stores facial images and feature values ​​of users who need to be closely monitored for illegal activities.

[0056] Edge Device: Terminal devices deployed in vehicles or on-site, responsible for real-time facial recognition and comparison.

[0057] Platform side (Vehicle Equipment Management System, VEMS): Central management platform, responsible for managing facial data, issuing feature values, receiving alarm information, etc.

[0058] Image transmission risks: During transmission, there is a security risk that the original facial images may be intercepted, leaked, or used illegally.

[0059] Network bandwidth consumption: Transmitting large amounts of image data consumes network bandwidth and affects the overall system performance.

[0060] Storage space consumption: The storage space occupied by storing a large number of original images on the device.

[0061] Encrypted transmission: Encryption algorithms are used to encrypt the transmitted facial images to ensure the security of the transmission process.

[0062] Access control: Through user permission management, control which users can view the original facial images.

[0063] Full update: The feature value index file on the device is completely replaced with new data, rather than being updated incrementally.

[0064] See Figure 1 This illustration shows a face recognition system provided in an embodiment of the present application. The face recognition system includes a feature value conversion device 1, a platform terminal 2, and a device terminal 3. The platform terminal is communicatively connected to the feature value conversion device and the device terminal, respectively.

[0065] A feature value conversion device is used to convert at least one face image obtained from the platform into corresponding feature values ​​and send them to the platform.

[0066] On the platform side, it is used to forward various feature values ​​to the device side;

[0067] On the device side, the acquired target face image is converted into target feature values, and the target feature values ​​are compared with each feature value. When the comparison result meets the preset conditions, a prompt message is triggered.

[0068] This application provides a face recognition system comprising a feature value conversion device, a platform, and a device. The feature value conversion device converts at least one face image acquired from the platform into feature values ​​and sends them to the platform. The platform forwards these feature values ​​to the device. The device converts the acquired target face image into target feature values ​​and compares these target feature values ​​with each of the other feature values. When the comparison results meet preset conditions, a notification message is triggered. Therefore, the technical solution provided in this application separates the feature value conversion task from the platform to an independent feature value conversion device. This not only decouples the system and reduces the computational burden on the platform, avoiding impact on its performance, but also ensures that face images are transmitted only between the platform and the feature value conversion device, thus avoiding the risk of data leakage at the device. Furthermore, the platform provided in this application only sends feature values ​​to the device, significantly reducing data transmission volume. The device provided in this application performs the corresponding feature comparisons, eliminating the need to store large numbers of face images, thus avoiding the consumption of significant storage space and network bandwidth, thereby improving face recognition efficiency.

[0069] Optionally, a feature value conversion device will be used to convert multiple face images obtained from the platform into feature values, specifically for:

[0070] The system periodically queries the face database on the platform to retrieve multiple face images from the platform; these face images are encrypted.

[0071] Each face image is converted into a feature value using a preset face recognition algorithm. However, the face image cannot be reconstructed using the feature value.

[0072] Combination Figure 1 See Figure 2 The feature value conversion device provided in this application includes a first smart camera and a first vehicle-mounted recording device. The device side includes a second smart camera, a second vehicle-mounted recording device, and a display screen. The smart camera can be a C27, which has corresponding AI algorithm capabilities and can convert face images into feature values. The smart camera can be either the first smart camera or the second smart camera. The vehicle-mounted recording device can be a V3N, which is used to connect to the smart camera and serves as a communication bridge between the smart camera and the platform. The vehicle-mounted recording device can be either the first vehicle-mounted recording device or the second vehicle-mounted recording device.

[0073] It should be noted that the display screen is used to show the face images that match the feature values ​​indicated in the face comparison results, which are fed back from the platform.

[0074] In some embodiments, the first vehicle-mounted video recording device periodically queries the face database task on the platform to obtain at least one face image from the platform based on the face database task; wherein the face image is an encrypted face image; the first smart camera uses a preset face recognition algorithm to convert the face image into feature values, wherein the face image cannot be reconstructed using the feature values.

[0075] Optionally, the eigenvalue transformation device is also used for:

[0076] Construct corresponding vector indexes based on the identifiers of the face images;

[0077] Establish a mapping relationship between vector indices and feature values ​​of face images, and establish a mapping relationship between vector indices and identifiers of face images;

[0078] Generate a feature value index file based on the feature values ​​of face images and their mapping relationships, and generate a metadata file based on the identifiers of face images and their mapping relationships;

[0079] Send the feature value index file and metadata file to the platform.

[0080] Optionally, a device that converts the real-time captured target face image into target feature values ​​is used for:

[0081] Real-time acquisition of target facial images;

[0082] A preset face recognition algorithm is used to convert a target face image into target feature values; however, the target face image cannot be reconstructed using the target feature values.

[0083] In some embodiments, a target face image is captured in real time by a second smart camera; the target face image is converted into target feature values ​​by the second smart camera using a preset face recognition algorithm; wherein the target face image cannot be reconstructed using the target feature values.

[0084] Optionally, a device that performs face comparison between the target feature value and each feature value separately is used for:

[0085] Calculate the similarity between the target feature value and each feature value;

[0086] Determine whether there is a similarity greater than a preset threshold among the various similarity scores;

[0087] If it does not exist, generate a comparison result that does not meet the preset conditions;

[0088] If such a similarity exists, the maximum similarity is determined from the similarities greater than the preset threshold, and a comparison result that meets the preset conditions is generated based on the feature value corresponding to the maximum similarity.

[0089] Optional, platform-side, specifically used for:

[0090] The receiving device sends comparison results that meet preset conditions;

[0091] Obtain the target vector index corresponding to the feature value indicated by the alignment result from the feature value index file;

[0092] Retrieve the target identifier matching the target vector index from the metadata file;

[0093] Obtain face images that match the target identifier, wherein the face images that match the target identifier are the face images whose feature values ​​match the comparison results;

[0094] Send the face image that matches the target identifier to the device.

[0095] Optionally, on the platform side, it is also used for:

[0096] Receive image requests sent by the receiving device, wherein the image requests include user information of the user to be verified;

[0097] Based on user information, determine whether the user to be verified has permission to view facial images;

[0098] If it exists, execute the step of sending the face image that matches the target identifier to the device;

[0099] If the face image is not present, a message indicating that the user does not have permission to view it should be sent to the device.

[0100] Based on the face recognition system provided in the above embodiments of this application, correspondingly, the embodiments of this application provide a face recognition method, such as... Figure 3 As shown, this method is applied to the face recognition system provided in the above-described embodiments of this application, and the method specifically includes the following steps:

[0101] S301: The feature value conversion device converts at least one face image obtained from the platform into corresponding feature values ​​and sends them to the platform.

[0102] In this embodiment, the platform can obtain facial images and user information of users requiring close monitoring from the blacklist facial database in real time. The user information may include the user's corresponding user identifier, user notes, and the tagged organization, etc. The user identifier can also be considered as the identifier of the user's facial image. For each facial image, a corresponding encryption algorithm can be used to encrypt the image. Facial images without corresponding feature values ​​are set to a waiting state. The user identifier can be a user name or user ID.

[0103] It should be noted that encrypting facial images can ensure the security of data transmission.

[0104] During the specific execution of step S301, the feature value conversion device can periodically obtain at least one face image from the platform. The face image obtained at this time is an encrypted face image in a waiting state. For each face image, a preset face recognition algorithm is used to convert the face image into feature values ​​that cannot be used to reconstruct the face image. The feature values ​​of each face image are then sent to the platform.

[0105] In some embodiments, when a face image in a waiting state exists on the platform, the platform can create a corresponding face database task. For example, the face database task can be a PENDING task to be processed. This allows the feature conversion device WDB to send requests to the platform at fixed intervals (e.g., 30 seconds) to query whether a face database task exists on the platform. When a face database task is found to exist on the platform and its state is PENDING, the WDB retrieves at least one face image in a waiting state from the platform.

[0106] Optionally, a feature value conversion device is used to convert multiple face images obtained from the platform into feature values. Specifically, this device is used to: periodically query the face database task on the platform to obtain at least one face image from the platform based on the face database task; wherein the face image is an encrypted face image; and convert the face image into feature values ​​using a preset face recognition algorithm, wherein the face image cannot be reconstructed using the feature values.

[0107] In this embodiment, the feature value conversion device includes a first smart camera and a first vehicle-mounted recording device. The process of converting multiple face images obtained from the platform into feature values ​​can be as follows: the first vehicle-mounted recording device periodically queries the face database task on the platform to obtain multiple face images from the platform based on the face database task; wherein, the face images are encrypted face images; the first smart camera uses a preset face recognition algorithm to convert each face image into a feature value, wherein the face image cannot be reconstructed using the feature values.

[0108] It is worth noting that the feature value conversion device can use the image-to-feature-value interface to call the face recognition algorithm built into the first smart camera to convert the face image into feature values. If the call fails, i.e., the feature value conversion fails, the call count can be incremented by 1. If the number of calls after incrementing by 1 does not exceed the call threshold, the image-to-feature-value interface is called again to call the face recognition algorithm built into the first smart camera to convert the face image into feature values. If the number of calls after incrementing by 1 exceeds the call threshold, the feature value conversion result is set to feature value conversion failure, the corresponding failure reason is recorded, and the call count is reset to zero.

[0109] In some embodiments, after querying the face database task from the platform, the feature value conversion device can obtain the face image in the waiting state from the platform through the pagination API, and calculate the feature value of the face image through the face recognition algorithm (such as the IRSE50 model) built into the first smart camera to generate a numerical vector (feature value) that cannot be restored to a face image.

[0110] In this embodiment, after generating feature values ​​for a face image, the feature value conversion device can also construct a corresponding vector index based on the face image's identifier before sending the feature values ​​to the platform; establish a mapping relationship between the vector index and the feature values ​​of the face image, as well as a mapping relationship between the vector index and the face image's identifier; generate a feature value index file based on the face image's feature values ​​and their mapping relationship, and generate a metadata file based on the face image's identifier and its mapping relationship; and send the feature value index file and metadata file to the platform.

[0111] In some embodiments, after obtaining the feature values ​​of a face image, the feature value conversion device can obtain the identifier of the face image, construct a vector index of the face image identifier using the HNSW algorithm, and construct a mapping relationship between the vector index and the feature values ​​of the face image and the identifier of the face image, respectively, so as to further utilize the feature values ​​of the face image and its mapping relationship to generate a feature value index file (vector_db_index.bin), and generate a metadata file (vector_db_index.metainfo) based on the identifier of the face image and its mapping relationship; finally, the generated two files are sent to the platform.

[0112] It should be noted that vector_db_index.bin is like the main archive, storing all the actual feature values ​​and retrieval paths (vector indexes); vector_db_index.metainfo is like the archive index, which can quickly determine where the information of a specific person (the identifier of a face image) is stored.

[0113] It should also be noted that by establishing a mapping relationship between the identifier of a face image and its corresponding vector index, and a mapping relationship between the feature value of a face image and its corresponding vector index, this mapping relationship can be used to quickly find the vector index corresponding to the feature value, and then the corresponding identifier can be quickly found based on the vector index, thereby quickly finding the corresponding face image.

[0114] Furthermore, in this embodiment, after the feature value conversion device feeds back the feature values ​​of the face image to the platform, it can update the task status of the face recognition task on the platform to READY. When a face image in a waiting state exists on the platform, it can create a corresponding face database task and set the task status of the face database task to PENDING. This allows the feature value conversion device to determine whether the task status of the face database task is PENDING when it queries the platform. If it is, it retrieves the face image in a waiting state from the platform; otherwise, it queries the face database task again.

[0115] S302: The platform forwards each feature value to the device.

[0116] In this embodiment, the platform can be used to store face images, feature values, metadata, and the status of maintenance instructions; wherein, the face images can be face images in a blacklist database, and the feature values ​​are feature values ​​in a feature value index file issued by a feature value conversion device.

[0117] It should be noted that the status of the specified action can be WAITTING (waiting for delivery), Success (delivery successful), Failed (delivery failed), or Issuing (delivery in progress). Among them, Waiting for delivery means that the platform needs to send the corresponding feature value to 100 devices and needs to wait in a queue; Success means that the device has successfully received the feature value sent by the platform and replied with a received status; Failed means that the device response timed out or failed, or the platform is in a faulty state; Issuing means that the feature value is currently being sent to the device (generally lasting 1-3 seconds).

[0118] During the specific execution step S302, when the platform receives the feature value index file and metadata file sent by the feature value conversion device, it can forward the feature value index file and metadata file containing the feature values ​​of the individual face image to the device.

[0119] In some embodiments, after receiving the feature value index file and metadata file, the device can perform a full update using the feature value index file and metadata file. Specifically, after detecting the feature value index file and metadata file sent by the platform, the device can delete the previously stored old feature value index file and metadata file to store the latest feature value index file and metadata file, avoiding data inconsistency issues caused by incremental updates and reducing process complexity.

[0120] S303: The device converts the collected target face image into target feature values, and compares the target feature values ​​with each feature value for face recognition. When the comparison result meets the preset conditions, a prompt message is triggered.

[0121] During the specific execution step S303, the device can acquire target face images in real time; convert the target face images into target feature values ​​using a preset face recognition algorithm; if the target face image cannot be reconstructed using the target feature values; compare the target feature values ​​with each feature value individually; when the comparison result meets preset conditions, trigger a prompt message to indicate that there is a user whose face image matches the blacklist face database; the comparison result that meets the preset conditions can also be sent to the platform, so that the platform sends the face image that matches the comparison result to the device. The preset conditions indicate that there are feature values ​​among the feature values ​​whose similarity to the target feature value is greater than a preset threshold.

[0122] In some embodiments, the device can receive a feature value index file sent by the platform through a second vehicle-mounted recording device; capture the face image of the currently passing user in real time through a second camera (for ease of distinction, the currently captured face image is referred to as the target face image), and convert the target face image into target feature values ​​through a second smart camera to determine whether there is a feature value in the feature value index file that matches the target feature value; if it exists, generate a comparison result that meets preset conditions, wherein the comparison result includes the feature value that matches the target feature value; if it does not exist, generate a comparison result that indicates a failure of face comparison, that is, generate a comparison result that does not meet the preset conditions, and continue to capture the target face image.

[0123] In this embodiment, the device can acquire a target face image in real time using a second smart camera; the target face image is converted into target feature values ​​using a preset face recognition algorithm via the second smart camera; however, the target face image cannot be reconstructed using the target feature values.

[0124] In some embodiments, the device can use a second smart camera to call the corresponding algorithm interface to invoke a face recognition algorithm to convert the target face image into feature values; if the call fails, i.e., the feature value conversion fails, the corresponding failure reason is recorded. If the call succeeds, the corresponding target feature values ​​are obtained.

[0125] It is worth noting that the device can use the face recognition algorithm (such as the IRSE50 model) built into the second smart camera to calculate the feature value of the target face image and generate a numerical vector (target feature value) that cannot be restored to the target face image.

[0126] Optionally, the process of comparing the target feature value with each feature value on the device side can be specifically as follows: The device side that compares the target feature value with each feature value on the device side is specifically used to: calculate the similarity between the target feature value and each feature value; determine whether there is a similarity greater than a preset threshold among the similarities; if not, generate a comparison result that does not meet the preset conditions; if so, determine the maximum similarity from the similarities greater than the preset threshold, and generate a comparison result that meets the preset conditions based on the feature value corresponding to the maximum similarity.

[0127] It should be noted that if there is only one similarity greater than the preset threshold, then that similarity can be taken as the maximum similarity, and the comparison result that meets the preset conditions can be generated based on the feature value corresponding to that similarity.

[0128] In some embodiments, the device calculates the similarity between the target feature value and each feature value through a second smart camera; determines whether there is a similarity greater than a preset threshold among the various similarities; if not, generates a comparison result representing a failed face comparison, i.e., generates a comparison result that does not meet the preset conditions; if there is, determines the maximum similarity from the similarities greater than the preset threshold, and generates a comparison result representing a successful face comparison based on the feature value corresponding to the maximum similarity, i.e., generates a comparison result that meets the preset conditions, wherein the comparison result includes the feature value that matches the target feature value (the feature value corresponding to the maximum similarity).

[0129] In this embodiment of the application, when the device determines that the face comparison result indicates that the face comparison is successful, it triggers a prompt message and sends the comparison result that meets the preset conditions to the platform through the second vehicle-mounted video recording device; wherein, the prompt message is used to indicate that the current face comparison is successful.

[0130] Furthermore, in this embodiment of the application, when the device determines that the face comparison is successful, it can assume that a face image in the blacklist face database has been matched. At this time, an alarm can be triggered, and a corresponding pop-up prompt, voice prompt, or corresponding prompt information can be displayed on the screen. At the same time, the corresponding comparison result is fed back to the platform.

[0131] In some embodiments, after the device triggers an alarm (triggers a prompt message), it can output a corresponding prompt message to prompt the user to be verified (such as the current staff member) to trigger the corresponding image request. This allows the platform to perform permission verification on the user to be verified based on the image request sent by the device, ensuring that only authorized users can view the corresponding face image, and further preventing data leakage.

[0132] In this embodiment of the application, after receiving the comparison result that meets the preset conditions from the device, the platform can find the face image that matches the feature value indicated by the face comparison result from the stored face images according to the feature value, feature value index file and metadata file indicated in the comparison result, and then feed back the found face image to the device.

[0133] Optionally, the platform receives the comparison results sent by the device that meet preset conditions; obtains the target vector index corresponding to the feature value indicated by the comparison result from the feature value index file; obtains the target identifier matching the target vector index from the metadata file; obtains the face image matching the target identifier, wherein the face image matching the target identifier is the face image matching the feature value indicated in the comparison result; and sends the face image matching the target identifier to the device.

[0134] In this embodiment of the application, to further ensure data security, before the platform sends the found face image (the face image matching the target identifier) ​​to the device, it can first check whether there is an image request sent by the user to be verified through the device. If there is an image request, the platform can use the image request to verify the user's permissions, that is, determine whether the user to be verified has permission to view the face image. If the permission verification passes, that is, the user has permission to view the face image, the platform sends the found face image to the device; otherwise, it sends a message to the device that the user does not have permission to view the face image.

[0135] Optionally, before sending a face image that matches the feature value indicated in the face comparison result to the device, the platform can receive an image request sent by the device, wherein the image request includes the user information of the user to be verified; based on the user information, it determines whether the user to be verified has permission to view the face image; if so, it executes the step of sending the face image that matches the target identifier to the device; if not, it sends a message to the device indicating that it does not have permission to view the face image.

[0136] It should be noted that user information may include the user identifier (ID) of the user to be verified. The platform can determine whether the user to be verified has the corresponding permission to view the face image based on the user identifier.

[0137] In some embodiments, after receiving a facial image from the platform, the device displays the facial image on a screen so that the user to be verified can view the corresponding facial image.

[0138] It should be noted that the device does not store the original face image. When a user needs to view the image, they click the "Fetch Pic" button to send an image request to the platform. After receiving the image request, the platform verifies the corresponding permissions and returns the face image after successful verification. In other words, the face image is only for viewing and is not permanently stored, which can further reduce the memory space occupied on the device and avoid data leakage.

[0139] This application provides a face recognition method comprising a feature value conversion device, a platform, and a device. The feature value conversion device converts at least one face image acquired from the platform into feature values ​​and sends them to the platform. The platform forwards each feature value to the device. The device converts the acquired target face image into target feature values ​​and compares the target feature values ​​with each of the other feature values. When the comparison result meets preset conditions, a prompt message is triggered. Therefore, the technical solution provided in this application separates the feature value conversion task from the platform to an independent feature value conversion device. This not only decouples the system and reduces the computational pressure on the platform, avoiding impact on the platform's system performance, but also ensures that face images are transmitted only between the platform and the feature value conversion device, thus avoiding the risk of data leakage at the device. Furthermore, the platform provided in this application is only responsible for sending feature values ​​to the device, significantly reducing data transmission volume. The device provided in this application is responsible for performing the corresponding feature comparisons, eliminating the need to store large numbers of face images, thus avoiding the consumption of large amounts of storage space and network bandwidth, thereby improving face recognition efficiency.

[0140] This application also provides a storage medium storing program instructions, which, when loaded and executed by a processor, implement any of the above-described face recognition method embodiments.

[0141] This application also provides an electronic device, such as Figure 4 As shown, the device includes a processor 401 and a memory 402, which are connected via a bus. The memory stores program instructions. The processor calls the program instructions in the memory to execute any of the above-described face recognition method embodiments.

[0142] The processor mentioned in this article can be the terminal's CPU, an integrated MCU within the terminal, or a combination of a CPU and an MCU. Furthermore, the processor contains a kernel that retrieves the corresponding program from memory; one or more kernels can be configured.

[0143] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0144] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0145] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0146] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0147] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A face recognition system, characterized in that, The face recognition system includes a feature value conversion device, a platform, and a device, wherein the platform is communicatively connected to the feature value conversion device and the device. The feature value conversion device is used to convert at least one face image obtained from the platform into corresponding feature values ​​and send them to the platform. The platform is used to forward each of the feature values ​​to the device. The device is used to convert the acquired target face image into target feature values, and compare the target feature values ​​with each of the feature values ​​for face recognition. When the comparison result meets the preset conditions, a prompt message is triggered.

2. The face recognition system according to claim 1, characterized in that, The feature value conversion device, which converts multiple face images acquired from the platform into feature values, is specifically used for: The face database task on the platform is periodically queried to obtain at least one face image from the platform based on the face database task; wherein the face image is an encrypted face image; The face image is converted into feature values ​​using a preset face recognition algorithm, wherein the face image cannot be reconstructed using the feature values.

3. The face recognition system according to claim 1, characterized in that, The eigenvalue conversion device is also used for: Construct a corresponding vector index based on the identifier of the face image; Establish a mapping relationship between the vector index and the feature values ​​of the face image, and establish a mapping relationship between the vector index and the identifier of the face image; A feature value index file is generated based on the feature values ​​of the face image and their mapping relationship, and a metadata file is generated based on the identifier of the face image and its mapping relationship; The feature value index file and the metadata file are sent to the platform.

4. The face recognition system according to claim 1, characterized in that, The device that converts real-time captured target face images into target feature values ​​is specifically used for: Real-time acquisition of target facial images; The target face image is converted into target feature values ​​using a preset face recognition algorithm; wherein the target face image cannot be reconstructed using the target feature values.

5. The face recognition system according to claim 1, characterized in that, The device that performs face comparison between the target feature value and each of the individual feature values ​​is specifically used for: Calculate the similarity between the target feature value and each of the feature values; Determine whether any of the aforementioned similarities exceeds a preset threshold; If it does not exist, generate a comparison result that does not meet the preset conditions; If such a similarity exists, the maximum similarity is determined from the similarities greater than the preset threshold, and a comparison result that meets the preset conditions is generated based on the feature value corresponding to the maximum similarity.

6. The face recognition system according to claim 3, characterized in that, The platform is also used for: Receive the comparison results sent by the device that meet the preset conditions; Obtain the target vector index corresponding to the feature value indicated by the alignment result from the feature value index file; Obtain the target identifier that matches the target vector index from the metadata file; Obtain a face image that matches the target identifier, wherein the face image that matches the target identifier is a face image whose feature value matches the feature value indicated in the comparison result; The face image matching the target identifier is sent to the device.

7. The face recognition system according to claim 6, characterized in that, The platform is also used for: Receive an image request sent by the device, wherein the image request includes user information of the user to be verified; Based on the user information, determine whether the user to be verified has permission to view facial images; If it exists, perform the step of sending the face image that matches the target identifier to the device. If not, send a message to the device indicating that the user does not have permission to view the face image.

8. A face recognition method, characterized in that, The method, applied to the face recognition system according to any one of claims 1-7, comprises: The feature value conversion device converts at least one face image obtained from the platform into corresponding feature values ​​and sends them to the platform. The platform forwards each of the feature values ​​to the device. The device converts the captured target face image into target feature values, and compares the target feature values ​​with each of the feature values. When the comparison result meets the preset conditions, a prompt message is triggered.

9. An electronic device, characterized in that, include: A processor and a memory are connected via a bus; wherein the processor is used to call and execute a program stored in the memory; The memory is used to store a program for implementing the face recognition method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions for performing the face recognition method as described in any one of claims 1-7.

Images