Personnel registration method, device, computer equipment and storage medium

Abstract

The present invention discloses a personnel registration method, apparatus, computer equipment, and storage medium. The method obtains identity registration information of a target community; the identity registration information includes a personnel registration picture of at least one community member belonging to the target community; obtains a portrait picture taken within a preset geographical range of the target community; the portrait picture includes at least one photographed subject; performs feature comparison between the personnel registration picture and the portrait picture to obtain an image comparison result between the personnel registration picture and the portrait picture; obtains third-party registration information corresponding to the target community, and determines the community registration status of the photographed subject included in the portrait picture in the target community based on the image comparison result and the third-party registration information. The present invention improves the efficiency and accuracy of personnel registration.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a personnel registration method, device, computer equipment and storage medium. Background Art

[0002] With the rapid increase in population and urban migration, the population in the community often changes. Therefore, in order to improve the safety of the community, it is very important to manage and register community members in a timely manner.

[0003] In existing technologies, the management and control of community residents often involves registering their information through voluntary reporting by residents or through visits by community service personnel. However, residents are not very enthusiastic about voluntary reporting, and the collection of information through visits by community service personnel requires a lot of manpower and resources. Furthermore, the efficiency of registering community residents is low and registration errors are prone to occur, ultimately resulting in a low accuracy rate for the registered information corresponding to community residents. Summary of the Invention

[0004] The embodiments of the present invention provide a personnel registration method, apparatus, computer equipment and storage medium to solve the problem of low efficiency and low accuracy in identity registration of people within a community in the prior art.

[0005] A personnel registration method, comprising:

[0006] Acquire identity registration information of a target community; the identity registration information includes a registration picture of at least one community member belonging to the target community;

[0007] Acquire a portrait picture taken within a preset geographical range of the target community; the portrait picture includes at least one photographed subject;

[0008] Performing feature comparison between the personnel registration picture and the portrait picture to obtain an image comparison result between the personnel registration picture and the portrait picture;

[0009] The third-party registration information corresponding to the target community is obtained, and the community registration status of the subject included in the portrait image in the target community is determined based on the image comparison result and the third-party registration information.

[0010] A personnel registration device, comprising:

[0011] A registration information acquisition module is used to acquire identity registration information of a target community; the identity registration information includes a registration photo of at least one community member belonging to the target community;

[0012] An image acquisition module is configured to acquire a portrait image taken within a preset geographical range of the target community; the portrait image includes at least one photographed subject;

[0013] An image feature comparison module is used to perform feature comparison between the personnel registration picture and the portrait picture to obtain an image comparison result between the personnel registration picture and the portrait picture;

[0014] The registration status confirmation module is used to obtain third-party registration information corresponding to the target community and determine the community registration status corresponding to each of the photographed objects based on the image comparison result and the third-party registration information.

[0015] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned personnel registration method is implemented.

[0016] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the personnel registration method is implemented.

[0017] The above-mentioned personnel registration method, device, computer equipment and storage medium obtain a portrait picture taken within a preset geographical range of the target community, and perform feature comparison between the portrait picture and the personnel registration picture. In this way, it is possible to determine whether the subject in the portrait picture matches the community personnel of the target community based on the image comparison result obtained by the feature comparison, thereby improving the efficiency of matching the subject with the community personnel; in addition, preset third-party registration information is introduced, and the third-party registration information is used to verify whether the subject has registration information, thereby improving the accuracy of the verification of the subject, and finally determining the community registration status of the subject in the target community based on the image comparison result and the third-party registration information, thereby improving the efficiency and accuracy of personnel registration. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0019] Figure 1 This is a schematic diagram of an application environment of a personnel registration method according to an embodiment of the present invention;

[0020] Figure 2is a flow chart of a personnel registration method according to an embodiment of the present invention;

[0021] Figure 3 This is a principle block diagram of a personnel registration device according to one embodiment of the present invention;

[0022] Figure 4 FIG. 1 is a schematic diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0024] The personnel registration method provided by the embodiment of the present invention can be applied as follows: Figure 1 Specifically, the personnel registration method is applied in a personnel registration system, which includes the following: Figure 1 The client and server shown in the figure can be applied to either the server or the client. The client and server communicate via a network, addressing the low efficiency and accuracy of identity registration for community members in existing technologies. The client, also known as the user end, refers to the program that corresponds to the server and provides local services to clients. The client can be installed on, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented as a standalone server or a server cluster consisting of multiple servers.

[0025] In one embodiment, if Figure 2 As shown, a personnel registration method is provided, which is applied in Figure 1 The server in the example is used as an example, and the steps are as follows:

[0026] S10: Acquire identity registration information of the target community; the identity registration information includes a registration picture of at least one community member belonging to the target community.

[0027] It is understandable that the target community can be a residential area, a street community, or a rural town, etc. The identity registration information can be the identity information registered in the target community. For example, when the target community is a residential area, the identity registration information can be the identity information of each family member reported by each owner, etc. Community personnel are people with registration information in the target community, such as owners, tenants, etc. in the target community. Among them, a community member has a personnel registration picture, and the personnel registration picture is an image containing facial features that represent the community member, preferably a facial image of each community member. Furthermore, the identity registration information can be stored in the database of the target community, and then the identity registration information can be directly obtained from the database after obtaining the permission of the database.

[0028] S20: Obtain a portrait picture taken within a preset geographical range of the target community; the portrait picture contains at least one photographed subject.

[0029] It is understood that portrait images are images captured by cameras or other cameras within the preset geographic range of the target community, such as those captured by surveillance equipment at each entrance of the target community or by cameras in different units within the target community. These portrait images are generally stored in a database associated with all cameras within the target community, and thus portrait images can be directly obtained from the database associated with these cameras. The preset geographic range can be limited based on the specific scenario, for example, the preset geographic range can be the entire target community. The subject of the portrait is the person corresponding to the facial features contained in the portrait image. A portrait image may contain one or more facial features, and each person corresponding to a different facial feature can be determined as a subject of the portrait.

[0030] It should be noted that in the subsequent step S30, image feature comparison of the portrait is required. Therefore, the portrait images are pre-filtered. That is, after performing portrait recognition on the images captured by each camera from the database associated with the camera, images containing at least one portrait information (i.e., facial features) are recorded as portrait images, while images without any portrait information are not included in the portrait images. This can improve the efficiency of feature comparison. Furthermore, in this embodiment, the portrait images are differentiated according to the actual shooting time. For example, portrait images taken on the same day are grouped together and compared with the personnel registration images. In this way, it is possible to determine whether community members appear in the target community based on portrait images taken on different days.

[0031] S30: performing feature comparison between the personnel registration picture and the portrait picture to obtain an image comparison result between the personnel registration picture and the portrait picture.

[0032] It can be understood that the feature comparison in this embodiment is to determine whether there are features in the portrait image that are the same as the facial features of the community personnel in the personnel registration image, that is, to compare a personnel registration image with the portrait image one by one, and then determine whether the feature similarity between the facial features in the portrait image and the facial features in the personnel registration image meets the preset conditions. The preset condition can be whether the feature similarity between the facial features in the portrait image and the facial features in the personnel registration image is greater than a preset similarity threshold. The preset similarity threshold can be set according to the specific scenario. For example, the preset similarity threshold can be set to 90%, 95%, etc. The image comparison result includes a result indicating a successful comparison and a result indicating a failed comparison. Among them, when an image comparison result indicates a successful comparison, it means that the facial features of the subject in the portrait image match the facial features in the personnel registration image, which means that the subject is a registered community member in the target community. When an image comparison result indicates a comparison failure, that is, the facial features of the subject in the portrait image do not match the facial features in all the registered images of the persons, which means that the subject is not registered in the target community.

[0033] Specifically, after obtaining portrait pictures taken within the preset geographical range of the target community, the personnel registration pictures of each community member are compared with all the portrait pictures one by one to determine whether the facial features of the subject in the portrait picture match the facial features in the personnel registration picture, and the image comparison results are obtained. Then, based on the image comparison results, it can be determined whether the subject in the portrait picture is a registered community member in the target community.

[0034] S40: Obtaining preset third-party registration information corresponding to the target community, and determining the community registration status of the subject included in the portrait image in the target community based on the image comparison result and the third-party registration information.

[0035] It is understandable that the preset third-party registration information in this embodiment refers to the registration information in the public security system of the target community. It should be noted that this preset third-party registration information is different from the identity registration information of the target community. The identity registration information of the target community is the basic information of each household initially collected by the target community, while the preset third-party registration information refers to the information of residents (currently or previously residing in the target community) who have registered their residence with the public security department of the target community. For example, community residents with identity registration information in the target community may not be registered in the preset third-party registration information.

[0036] Furthermore, the community registration status refers to the community registration status of the subject included in the portrait image in the target community. Among them, the community registration status in this embodiment includes but is not limited to the state to be analyzed, the unregistered state, the registered state, the uncancelled state, the cancelled state, the verified state and the whitelist state. The subject in the state to be analyzed may be one for which the registration status analysis has not yet been performed, such as a subject that is photographed for the first time. A subject in a registered state means that the subject frequently appears in the target community, and the subject has been successfully registered in the identity registration information of the target community and the third-party registration information; a subject in an unregistered state means that the subject frequently appears in the target community, and the subject matches one of the community personnel, but the subject is not registered in the third-party registration information; a subject in an uncancelled state means that the subject does not appear in the target community or appears in the target community less frequently (such as only once a month), but the subject still has registration information in the third-party registration information; a subject in a cancelled state means that the subject does not appear in the target community or appears in the target community less frequently, and The third-party registration information indicates that the subject no longer resides or operates in the target community; the verified status means that the subject appears in the portrait photo but is not registered in the identity registration information of the target community, nor is it registered in the third-party registration information. In this case, it is necessary to automatically identify the subject through the image features of the subject and determine its identity through manual verification; the whitelist status refers to the subject who frequently appears in the target community but is not a resident of the target community, and the subject is not registered in the third-party registration information, such as deliverymen, community security guards, community property, etc. In this way, this type of subject can be classified as a whitelist person, that is, when such a person is identified, his community registration status can be classified as a whitelist status.

[0037] Specifically, after performing feature comparison between the personnel registration picture and the portrait picture and obtaining the image comparison result between the personnel registration picture and the portrait picture, the preset third-party registration information of the public security system of the jurisdiction of the target community is obtained. Then, based on the image comparison result and the preset third-party registration information, it can be determined whether the photographed subject matches the community personnel and whether the identity registration is carried out in the public security system of the jurisdiction of the target community, and then the community registration status of the photographed subject contained in the portrait picture in the target community can be determined.

[0038] In this embodiment, by obtaining a portrait picture taken within a preset geographical range of the target community, and performing feature comparison between the portrait picture and the personnel registration picture, it is possible to determine whether the subject in the portrait picture matches the community personnel of the target community based on the image comparison result obtained by the feature comparison, thereby improving the efficiency of matching the subject with the community personnel; in addition, preset third-party registration information is introduced, and the third-party registration information is used to verify whether the subject has registration information, which can improve the accuracy of the verification of the subject, and finally determine the community registration status of the subject in the target community based on the image comparison result and the third-party registration information, thereby improving the efficiency and accuracy of personnel registration.

[0039] In one embodiment, in step S20, feature comparison is performed between the person registration picture and the portrait picture to obtain an image comparison result between the person registration picture and the portrait picture, including:

[0040] The clarity of the portrait image is obtained, and a clear portrait image that meets a preset clarity condition is generated based on the clarity of the portrait image.

[0041] Understandably, the clarity of portrait images can be easily affected by cameras or other camera equipment, which can cause them to appear blurry. Therefore, it is necessary to first perform a clarity test on the portrait images. For example, the clarity value of each captured portrait image can be determined using a Laplacian gradient function, SMD2 (grayscale variance product) function, NRSS gradient structural similarity method, etc. In this embodiment, a preset clarity threshold and the clarity of each portrait image are used to determine whether each portrait image meets the preset clarity condition. The preset clarity threshold can be determined based on the specific application scenario. For example, the preset clarity threshold can be set to 90%, 95%, etc. (90%, 95% refers to the clarity of the portrait image).

[0042] The clear portrait picture is compared with the personnel registration picture to obtain an image comparison result between the personnel registration picture and the clear portrait picture.

[0043] Specifically, after obtaining the clarity of the portrait picture and generating a clear portrait picture that meets the preset clarity conditions based on the clarity of the portrait picture, it can be ensured that the clarity of all pictures that are feature-matched with the personnel registration picture can meet the preset clarity conditions, and then by performing feature comparison between the clear portrait picture and the personnel registration picture, the image comparison result between the personnel registration picture and the clear portrait picture can be determined.

[0044] In this embodiment, the clarity of the portrait pictures is adjusted by setting a preset clarity condition, so that the clarity of all pictures used for feature comparison with the personnel registration pictures can meet the preset clarity condition, thereby improving the accuracy of picture feature comparison.

[0045] In one embodiment, generating a clear portrait image that meets a preset clarity condition based on the clarity of the portrait image includes:

[0046] Compares the clarity of the portrait image to a preset clarity threshold.

[0047] Specifically, after obtaining the clarity of the portrait picture, the clarity of the portrait picture is compared with a preset clarity threshold, so as to determine whether the portrait picture meets the preset clarity condition.

[0048] If the clarity of the portrait picture is less than a preset clarity threshold, the portrait picture is input into a preset image conversion model to generate a portrait clear picture corresponding to the portrait picture through the preset image conversion model; the clarity of the portrait clear picture is greater than or equal to the preset clarity threshold.

[0049] It can be understood that the preset image conversion model proposed in this embodiment is constructed based on two bidirectional adversarial networks, and the preset image conversion model can convert a picture with lower clarity into a picture with higher clarity. Among them, the preset image conversion model includes a first image conversion network and a second image conversion network, and the first image conversion network and the second image conversion network both contain the same first generator and second generator. Furthermore, the first generator is used to convert a blurred image into a clear image, and the second generator is used to convert a clear image into a blurred image. Therefore, this embodiment adopts the first generator in the preset image conversion model to convert a portrait picture that does not meet the preset clarity conditions into a portrait clear picture that meets the preset clarity conditions.

[0050] Specifically, after comparing the clarity of the portrait image with a preset clarity threshold, if the clarity of the portrait image is less than the preset clarity threshold, it indicates that the portrait image does not meet the preset clarity condition. The portrait image is then input into a preset image conversion model to generate a clear portrait image corresponding to the portrait image using the preset image conversion model. The generated clear portrait image is greater than or equal to the preset clarity threshold. Thus, after image conversion using the preset image conversion model, the portrait image that does not meet the preset clarity condition can be converted into a clear portrait image that meets the preset clarity condition.

[0051] If the portrait image is greater than or equal to the preset clarity threshold, the portrait image is directly recorded as a portrait clarity image.

[0052] Specifically, after comparing the clarity of the portrait image with a preset clarity threshold, if the clarity of the portrait image is greater than or equal to the preset clarity threshold, it indicates that the portrait image meets the preset clarity condition, and the portrait image is directly recorded as a portrait clarity image without the need for clarity optimization through a preset image conversion model.

[0053] In this embodiment, a preset image conversion model is introduced to adjust the clarity of portrait pictures that do not meet the preset clarity conditions, thereby generating clear portrait pictures that meet the preset clarity conditions. This improves the clarity of the pictures while providing a high-quality picture basis for feature comparison between the portrait pictures and the personnel registration pictures in the subsequent steps, thereby improving the accuracy of feature comparison.

[0054] In one embodiment, before inputting the portrait image into the preset image conversion model, the method further includes:

[0055] A sample image set is obtained; the sample image set includes at least one sample blurred image and a sample clear image corresponding to the sample blurred image.

[0056] It is understandable that the sample blurred image is an image of lower quality (e.g., blurred). The sample blurred image can be obtained by shooting with a handheld camera (e.g., when the hand is shaking, the shooting focus is inaccurate, or the pixel of the shooting device is low), or it can be obtained by processing the sample clear image such as blurring and pixel adjustment. The sample clear image can be a normal shot with high clarity, or a high-definition image crawled using crawler technology. It is understandable that the difference between the sample blurred image and its corresponding sample clear image is only the image clarity, and the essential content of the images is the same.

[0057] A preset recognition model including initial parameters is obtained; the preset recognition model includes a first image conversion network and a second image conversion network.

[0058] It can be understood that the preset recognition model includes a first image conversion network and a second image conversion network. Among them, the first image conversion network and the second image conversion network are both adversarial generative networks, with two generators and two discriminators in the first image conversion network and two generators and two discriminators in the second image conversion network. The generators and discriminators in the first image conversion network and the second image conversion network are synchronized, that is, when the initial parameters of the preset recognition model are adjusted and updated, the generators and discriminators in the first image conversion network and the second image conversion network are updated synchronously and the parameters are the same.

[0059] Furthermore, the first generator in the first image conversion network and the second image conversion network is used to convert a blurred image into a clear image, and the second generator is used to convert a clear image into a blurred image; in the first image conversion network, the first discriminator associated with the first generator is used to determine whether the image generated by the first generator is the same as the sample clear image; and the second discriminator associated with the second generator is used to determine whether the image generated by the second generator is the same as the sample blurred image. In the second image conversion network, the second discriminator associated with the second generator is used to determine whether the image generated by the second generator based on the sample clear image is the same as the sample blurred image; and the first discriminator associated with the first generator is used to determine whether the image generated by the first generator is the same as the sample clear image.

[0060] The sample blurred image is input into the first image conversion network to obtain a predicted blurred image corresponding to the sample blurred image; the sample clear image is input into the second image conversion network to obtain a predicted clear image corresponding to the sample clear image.

[0061] Specifically, after the sample blurred image is input into the first image conversion network, the sample blurred image is converted into an adversarial clear image by the first generator in the first image conversion network, and the adversarial clear image is converted into a predicted blurred image by the second generator in the first image conversion network; after the sample clear image is input into the second image conversion network, the sample clear image is converted into an adversarial blurred image by the second generator in the second image conversion network, and the adversarial blurred image is converted into a predicted clear image by the first generator in the second image conversion network.

[0062] A first loss parameter of a first image conversion network is determined according to the sample blurred image and the predicted blurred image; a second loss parameter of a second image conversion network is determined according to the sample clear image and the predicted clear image.

[0063] It can be understood that the first loss parameter and the second loss parameter are used to reflect the errors that occur during the image conversion process of the preset recognition model. Since the preset recognition model may have image conversion errors before training, the preset recognition model is continuously adjusted through the first loss parameter and the second loss parameter to improve the image conversion accuracy of the preset recognition model.

[0064] Among them, the first loss parameter includes the adversarial network loss value and the image feature loss value. The adversarial network loss value includes the loss values ​​of the first generator and the second generator. In the above description, it is pointed out that in the first image conversion network, the sample blurred image will be converted into an adversarial clear image and then into a predicted blurred image. Therefore, the adversarial network loss value in the first loss parameter is the sum of the loss between the sample blurred image and the adversarial clear image, and the loss between the adversarial clear image and the predicted blurred image. The image feature loss value can be determined by the feature similarity between the sample blurred image and the predicted blurred image. Furthermore, the loss between the sample blurred image and the adversarial clear image can be determined by the first discriminator corresponding to the first generator; the loss between the adversarial clear image and the predicted blurred image can be determined by the second discriminator corresponding to the second generator.

[0065] Among them, the second loss parameter also includes the adversarial network loss value and the image feature loss value. The adversarial network loss value includes the loss values ​​of the first generator and the second generator. In the above description, it is pointed out that the sample clear image will be converted into an adversarial blurred image and then into a predicted clear image in the second image conversion network. Therefore, the adversarial network loss value in the second loss parameter is the loss between the sample clear image and the adversarial blurred image, and the sum of the losses between the adversarial blurred image and the predicted clear image. The image feature loss value can be determined by the feature similarity between the sample clear image and the predicted clear image. Furthermore, the loss between the sample clear image and the adversarial blurred image can be determined by the second discriminator corresponding to the second generator; the loss between the adversarial blurred image and the predicted clear image can be determined by the first discriminator corresponding to the first generator.

[0066] Furthermore, although the first generator and the second generator in the first image conversion network and the second image conversion network are the same, since the input to the first image conversion network is a blurred sample image and the input to the second image conversion network is a clear sample image, the loss values ​​of the first image conversion network and the second image conversion network are different.

[0067] A predicted loss value of a preset recognition model is determined based on the first loss parameter and the second loss parameter.

[0068] Specifically, after determining the first loss parameter of the first image conversion network based on the sample blurred image and the predicted blurred image; and determining the second loss parameter of the second image conversion network based on the sample clear image and the predicted clear image, the first loss parameter and the second loss parameter can be linearly superimposed to obtain the predicted loss value of the preset recognition model.

[0069] When the predicted loss value does not reach the preset convergence condition, the initial parameters in the preset recognition model are iteratively updated until the predicted loss value reaches the convergence condition, and the preset recognition model after convergence is recorded as the preset image conversion model.

[0070] It can be understood that the convergence condition can be the condition that the predicted loss value is less than the set threshold, that is, when the predicted loss value is less than the set threshold, the training is stopped; the convergence condition can also be the condition that the predicted loss value is very small after 10,000 calculations and will not decrease anymore, that is, when the predicted loss value is very small after 10,000 calculations and will not decrease, the training is stopped and the preset recognition model after convergence is recorded as the preset image conversion model.

[0071] Furthermore, after determining the predicted loss value of the preset recognition model based on the first loss parameter and the second loss parameter, when the predicted loss value does not reach the preset convergence condition, the initial parameters of the preset recognition model are adjusted according to the predicted loss value, and the sample blurred image and the corresponding sample clear image are re-input into the preset recognition model after adjusting the initial parameters, so that when the predicted loss value of the sample blurred image reaches the preset convergence condition, another sample blurred image in the sample image set is selected, and the above steps are performed to obtain the predicted loss value corresponding to the sample blurred image, and when the predicted loss value does not reach the preset convergence condition, the initial parameters of the preset recognition model are adjusted again according to the predicted loss value, so that the predicted loss value of the sample blurred image reaches the preset convergence condition.

[0072] In this way, after the preset recognition model is trained through all the sample blurred images in the sample image set, the results output by the preset recognition model can continue to approach the accurate results, making the recognition accuracy higher and higher, until the prediction loss values ​​of all sample blurred images reach the preset convergence conditions, and the preset recognition model after convergence is recorded as the preset image conversion model.

[0073] In this embodiment, a preset recognition model including two image conversion networks is introduced, and it is trained using a number of sample blurred images. In addition, an adversarial loss value and an image feature loss value are added to the loss of the preset recognition model. This improves the efficiency of training the preset recognition model, and also improves the accuracy of the image conversion model obtained after training.

[0074] In one embodiment, a feature comparison is performed on the clear portrait image and the person registration image to obtain an image comparison result between the person registration image and the clear portrait image, including:

[0075] Performing portrait recognition on the clear portrait picture to obtain at least one captured portrait picture; the captured portrait picture refers to a portrait picture captured from the clear portrait picture and containing only one subject.

[0076] The captured portrait images containing the same subject are associated and recorded as a captured portrait group.

[0077] It can be understood that portrait recognition is a method for identifying different subjects in a clear portrait picture, and then different facial features in the clear portrait picture can be classified into a portrait group of the corresponding subject. Among them, one subject corresponds to a portrait group, and a portrait group includes a portrait picture of the subject captured from the clear portrait picture, that is, a captured portrait picture; a portrait group includes at least one captured portrait picture. For example, when the subject exists in different clear portrait pictures, the portrait picture of the subject can be captured from different clear portrait pictures. In this way, after the clear portrait pictures are clustered and the portrait groups are obtained, the efficiency of feature comparison between the personnel registration pictures and the clear portrait pictures in the subsequent steps can be improved.

[0078] For each portrait group, the image similarity between the person registration image and the captured portrait image in the portrait group is determined, and the image comparison result between the person registration image and the clear portrait image is determined based on the image similarity.

[0079] Specifically, after performing portrait clustering on the clear portrait pictures and obtaining at least one portrait group containing the cropped portrait pictures of the same photographed subject, the person registration picture can be compared with the cropped portrait pictures in each portrait group to determine the image similarity between the person registration picture and the cropped portrait pictures in each portrait group, and then determine the image comparison result between the person registration picture and the clear portrait picture based on the image similarity.

[0080] In this embodiment, after obtaining a captured portrait group by performing portrait clustering on the clear portrait pictures, the person registration picture can be compared with the cropped portrait pictures in the captured portrait group, rather than comparing the person registration picture with any clear portrait picture. Since the facial features of the same subject are the same, comparing the person registration picture with the cropped portrait pictures of the same subject can better notice the differences between the facial features, thereby improving the efficiency and accuracy of feature comparison.

[0081] In one embodiment, the image comparison result is associated with the shooting time of the portrait image; and determining the community registration status of the subject included in the portrait image in the target community based on the image comparison result and the third-party registration information includes:

[0082] Obtaining a preset time range, and determining at least one historical shooting time according to the shooting time of the portrait picture and the preset time range; the historical shooting time is earlier than the shooting time of the portrait picture;

[0083] It is understood that the preset time range can be set as needed, for example, the preset time range is set to the first 6 days before the shooting time. The historical shooting time is earlier than the shooting time of the portrait shooting time picture. It is understandable that, assuming that the preset time range is set to the first 6 days before the shooting time, the historical shooting time can be one day before the shooting time, two days before the shooting time, etc.

[0084] A historically captured picture corresponding to a historical capturing time is obtained; the historically captured picture includes at least one captured object.

[0085] Understandably, historical images are images taken at historical times within the preset geographic area of ​​the target community. A group of historical images is associated with each historical time, meaning that historical images taken at the same time are grouped together. Similar to portrait images, historical images still contain at least one subject, which may or may not have appeared in the portrait image.

[0086] According to the historical pictures and the portrait pictures, portrait tracking is performed on the subjects included in the historical pictures to obtain portrait tracking results corresponding to the subjects included in the historical pictures.

[0087] It can be understood that the portrait tracking result represents the number of times the subject appears in the historical pictures and the portrait pictures (in this embodiment, the unit is day. If the subject appears multiple times in a day, it is also counted as one time. For example, if there are multiple portrait pictures containing the subject, it is determined that the subject appears once in the shooting time of the portrait pictures; if a subject appears in the portrait picture but does not appear in the historical pictures taken at one of the historical shooting times, it is determined that the subject did not appear in the historical shooting time). Specifically, after obtaining the historical pictures corresponding to the historical shooting time, portrait tracking is performed on the subjects included in the historical pictures based on the historical pictures and the portrait pictures. The number of times each subject included in the historical pictures appears in the pictures taken at different shooting times can be determined, and then the portrait tracking results corresponding to each subject can be obtained.

[0088] For example, suppose a subject appears in the portrait subject, but does not appear in any of the historical images taken at other historical shooting times. If the portrait tracking condition is set to require the subject to appear on at least three days within seven days, then the portrait tracking result corresponding to the subject indicates a tracking failure. Suppose a subject appears in the portrait subject, but appears in four of the historical images taken at other historical shooting times and does not appear in two of the historical images taken at other historical shooting times. If the portrait tracking condition is set to require the subject to appear on at least three days within seven days, then the portrait tracking result corresponding to the subject indicates a tracking success.

[0089] Based on the portrait tracking results, image comparison results and third-party registration information, the community registration status of the subject included in the portrait image in the target community is determined.

[0090] Specifically, by performing portrait tracking on each subject based on historical photos and portrait photos, and obtaining portrait tracking results corresponding to each subject, the community registration status corresponding to the subject can be determined based on third-party registration information, portrait tracking results corresponding to the same subject, and image comparison results.

[0091] In this embodiment, the portrait tracking results obtained by portrait tracking the subject are determined, and the community registration status corresponding to each subject is determined through three aspects: the portrait tracking results, the image comparison results, and the third-party registration information. This allows the community registration status of each subject to be more accurately distinguished, thereby improving the accuracy and efficiency of community registration status classification.

[0092] In one embodiment, the community registration status includes: pending analysis status, unregistered status, registered status, unregistered status, and cancelled status; the third-party registration information includes a cancelled person information table and an existing person information table;

[0093] Based on the portrait tracking results, image comparison results, and third-party registration information, determine the community registration status of the subject in the portrait image in the target community, including:

[0094] According to the portrait picture, the deregistered personnel information table and the existing personnel information table, the identity registration result of the photographed subject included in the portrait picture is determined.

[0095] It can be understood that the deregistered personnel information table refers to the information of people who have been deregistered from the public security system of the target community, such as people who originally lived in the target community but have moved out. This deregistered personnel information table includes the deregistered portrait images of each deregistered person. The existing personnel information table refers to the information of people currently registered with the public security system of the target community, such as people who still live in the target community. This existing personnel information table includes the current portrait images of each existing person.

[0096] Specifically, the deregistered personnel information table and the existing personnel information table in the third-party registration information are obtained, and the facial features of each subject in the portrait image are compared with the facial features of each deregistered portrait image and the facial features of the existing portrait image to determine whether the subject is registered in the deregistered personnel information table or the existing personnel information table, thereby obtaining an identity registration result. Among these, the identity registration result may fall into the following three situations: the facial features of the subject successfully match only with the facial features of the deregistered portrait image in the deregistered personnel information table; the facial features of the subject successfully match only with the facial features of the existing portrait image in the existing personnel information table; or the facial features of the subject fail to match both with the facial features of the deregistered portrait image in the deregistered personnel information table and with the facial features of the existing portrait image in the existing personnel information table. If the facial features of the subject successfully match both with the facial features of the deregistered portrait image in the deregistered personnel information table and with the facial features of the existing portrait image in the existing personnel information table, an exception is output to check whether there are any registration errors in the deregistered personnel information table and the existing personnel information table.

[0097] When the portrait tracking result indicates that tracking has failed, the image comparison result indicates that comparison has been successful, and the identity registration result indicates that the portrait photo has successfully matched the existing personnel information table but failed to match the cancelled personnel information table, the community registration status is determined to be the non-cancelled status.

[0098] It can be understood that when the portrait tracking result indicates tracking failure, it indicates that the community member appears in the target community for fewer days within the preset time range. The image comparison result indicates successful comparison, indicating that the subject is registered in the target community, that is, the subject may be an existing resident or a historical resident of the target community. If the identity registration result of the subject indicates that the portrait photo successfully matches the existing personnel information table but fails to match the cancelled personnel information table, it indicates that the subject no longer lives in the target community but still has a record in the existing personnel information table of the preset third-party registration information. Therefore, the community registration status of the community member is determined to be an uncancelled status. In this way, the subject's identity information can be removed from the existing personnel information table and added to the cancelled personnel information table.

[0099] When the portrait tracking result indicates that tracking has failed, the image comparison result indicates that comparison has been successful, and the identity registration result indicates that the portrait photo has failed to match the existing personnel information table but has successfully matched the cancelled personnel information table, the community registration status is determined to be cancelled.

[0100] It can be understood that when the portrait tracking result indicates tracking failure, it indicates that the subject appears in the target community for fewer days within the preset time range. The image comparison result indicates successful comparison, which indicates that the subject is registered in the target community, that is, the subject may be an existing resident or a historical resident of the target community. If the identity registration result of the subject indicates that the portrait picture fails to match the existing personnel information table, but successfully matches the cancelled personnel information table, it indicates that the subject no longer lives in the target community and there is no information record in the existing personnel information table, but there is an information record in the cancelled personnel information table, which means that the subject has been cancelled in the third-party registration information, and therefore its community registration status is determined to be cancelled.

[0101] When the portrait tracking result indicates that the tracking is successful, the image comparison result indicates that the comparison is successful, and the identity registration result indicates that the person's photo fails to match the existing person information table and the cancelled person information table, the community registration status is determined to be unregistered.

[0102] It can be understood that when the portrait tracking result indicates successful tracking, it indicates that the subject has appeared in the target community for more days within the preset time range. The image comparison result indicates successful comparison, which indicates that the subject is registered in the target community, that is, the subject may be an existing resident or a historical resident of the target community. If the identity registration result of the subject indicates that the person's photo fails to match the existing personnel information table and the cancelled personnel information table, it indicates that the subject may live in the target community, but is not registered in the existing personnel information table of the preset third-party registration information. Therefore, its community registration status is recorded as unregistered, and the subject's personnel registration information is added to the existing personnel information table.

[0103] When the portrait tracking result indicates that the tracking is successful, the image comparison result indicates that the comparison is successful, and the identity registration result indicates that the person's photo is successfully matched with the existing person information table but fails to match with the deregistered person information table, the community registration status is determined to be registered.

[0104] It can be understood that when the portrait tracking result indicates successful tracking, it indicates that the subject has appeared in the target community for more days within the preset time range. The image comparison result indicates that the comparison is successful, which indicates that the subject is registered in the target community, that is, the subject may be an existing resident or a historical resident of the target community. If the identity registration result of the subject indicates that the person's photo is successfully matched with the existing person information table and fails to match with the deregistered person information table, it indicates that the subject has been registered in the existing person information table, and its community registration status is recorded as registered.

[0105] In this embodiment, by distinguishing portrait tracking results, image comparison results, and identity registration results of people with different community registration statuses, the accuracy of classifying the community registration status of each photographed subject can be increased.

[0106] In one embodiment, the image comparison result is associated with the shooting time of the portrait image; after determining the community registration status of the subject included in the portrait image in the target community, the following is further included:

[0107] The third-party registration information is updated according to the community registration status corresponding to each photographed object to obtain updated third-party registration information.

[0108] It can be understood that after determining the community registration status of the subjects included in the portrait image in the target community, the preset third-party registration information can be updated according to the community registration status corresponding to each subject. For example, when the community registration status is unregistered, the personal identity information of the unregistered subject can be added to the existing personnel information in the preset third-party registration information, thereby obtaining updated third-party registration information.

[0109] Acquire updated pictures that are within a preset geographic range of the target community and after the time at which the portrait picture was taken.

[0110] It can be understood that the updated captured picture is a picture captured by a camera or other camera equipment in the target community after the capturing time of the portrait captured picture.

[0111] Perform feature comparison on the personnel registration picture and the updated picture to obtain an updated image comparison result between the personnel registration picture and the updated picture.

[0112] Specifically, after obtaining an updated image captured within a preset geographic range of the target community and after the time the portrait image was captured, a feature comparison is performed between the person registration image and the updated image to obtain an updated image comparison result between the person registration image and the updated image. The method for performing feature comparison between the person registration image and the updated image is the same as the method for performing feature comparison between the person registration image and the portrait image described above, and will not be repeated here.

[0113] Acquire the registration status transition condition corresponding to the community registration status corresponding to each photographed object.

[0114] It can be understood that the registration status transition conditions are the conditions for transitioning between community registration states. For example, assuming the subject's community registration status is unregistered, the registration transition condition for transitioning from unregistered to registered is that the subject's personal identity information is registered in the existing personal information table within the preset third-party registration information. Assuming the subject's community registration status is not cancelled, the registration transition condition for transitioning from cancelled to registered is that the subject's portrait tracking result indicates successful tracking.

[0115] An updated registration state corresponding to each photographed object is determined according to the updated image comparison result, the updated third-party registration information, and the registration state transition condition.

[0116] Specifically, after obtaining the registration status conversion conditions corresponding to the community registration status corresponding to each photographed object, it can be determined whether the photographed object meets the registration status conversion conditions corresponding to its current community registration status based on the updated image comparison results, the updated third-party registration information and the registration status conversion conditions. If the registration status conversion conditions are met, its current community registration status is updated to the community registration status after the registration conversion conditions, that is, the updated registration status.

[0117] In this embodiment, the updated captured image collected in real time after the shooting time is compared with the personnel registration image to determine the updated image comparison result, and then, based on the updated image comparison result, the updated third-party registration information and the registration status conditions, it can be determined whether the community registration status corresponding to the photographed object needs to be updated. Then, if it is determined that the registration status conversion conditions are met, its community registration status is updated. If the registration status conversion conditions are not met, the current community registration status is retained. In this way, the community registration status of the photographed object can be changed in real time, thereby improving the flexibility and efficiency of changing the community registration status.

[0118] It should be noted that the above steps are to confirm the community registration status of the subjects included in the portrait images. In addition, the present invention can also perform the following activity status analysis for each subject:

[0119] First: by obtaining the time when the subject first appears in the portrait pictures (such as the above-mentioned portrait pictures or updated pictures) of each day, and the time when the subject last appears. For example, when performing feature comparison between the personnel registration pictures and the portrait pictures in the above steps, the portrait picture with the earliest shooting time in the image comparison results representing successful comparison is determined as the time when the subject first appears, and the portrait picture with the latest shooting time in the image comparison results representing successful comparison is determined as the time when the subject last appears; then, by comparing the first appearance time with a set early exit time threshold (such as 6 o'clock), and the last appearance time with a set late return time threshold (such as 11 o'clock); if the first appearance time is earlier than the set early exit time threshold and the last appearance time is later than the set late return time threshold, further use the portrait pictures to assist in determining whether the subject entered the target community from outside the target community, or whether the subject went from outside the target community to outside the target community; if the subject went from outside the target community to outside the target community at the first appearance time and went from outside the target community to inside the target community at the last appearance time, then the subject's activity status is confirmed to be early exit and late return.

[0120] Second: If the above steps determine that the first appearance time is earlier than the set early exit time threshold, and the last appearance time is later than the set late return time threshold, and the portrait picture is used to assist in determining that the subject's first appearance time is from outside the target community to entering the target community, and the last appearance time is from inside the target community to outside the target community, then it is confirmed that the subject's activity status is nocturnal.

[0121] Third: If a subject has multiple successful image comparison results in the portrait images taken on one day, and the subject's stay time in the target community is greater than the set minimum stay time (such as one hour) and less than the set maximum stay time (such as three hours, etc.), then the subject can be marked as a suspicious person, and the image features of the suspicious person can be transmitted to the criminal recognition model in the public security system, providing a data basis for the criminal recognition model.

[0122] Fourth: In the identity registration information, the age information of each community member can be identified, and then the community members whose age is older than the set threshold (such as 60 years old) can be marked as special concern persons, and then the time of appearance of the special concern person in the target community can be obtained through portrait photography. If the special concern person does not appear in the target community for more than the set time threshold (such as more than 5 hours since the last appearance in the target community), a suspected missing person alert will be sent to the mobile terminal of the community personnel associated with the special concern person (such as the children of the special concern person, etc.) (the mobile terminal can obtain the contact information of the community personnel associated with the special concern person through the identity registration information).

[0123] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0124] In one embodiment, a personnel registration device is provided, which corresponds to the personnel registration method in the above embodiment. Figure 3 As shown, the personnel registration device includes a registration information acquisition module 10, an image acquisition module 20, an image feature comparison module 30, and a registration status confirmation module 30. The functional modules are described in detail as follows: the registration information acquisition module 10 is used to obtain the identity registration information of the target community; the identity registration information includes a personnel registration picture of at least one community member belonging to the target community; the image acquisition module 20 is used to obtain a portrait picture taken within a preset geographical range of the target community; the portrait picture includes at least one photographed subject; the image feature comparison module 30 is used to perform feature comparison between the personnel registration picture and the portrait picture to obtain an image comparison result between the personnel registration picture and the portrait picture; the registration status confirmation module 40 is used to obtain third-party registration information corresponding to the target community, and determine the community registration status corresponding to each photographed subject based on the image comparison result and the third-party registration information.

[0125] Preferably, the image feature comparison module 30 includes: a clarity acquisition unit, used to acquire the clarity of the portrait picture, and generate a clear portrait picture that meets the preset clarity conditions based on the clarity of the portrait picture; a feature comparison unit, used to perform feature comparison between the clear portrait picture and the personnel registration picture, and obtain an image comparison result between the personnel registration picture and the clear portrait picture.

[0126] Preferably, the clarity acquisition unit includes: a clarity comparison subunit, used to compare the clarity of the portrait picture with a preset clarity threshold; a picture conversion subunit, used to input the portrait picture into a preset image conversion model if the clarity of the portrait picture is less than the preset clarity threshold, so as to generate a clear portrait picture corresponding to the portrait picture through the preset image conversion model; the clarity of the clear portrait picture is greater than or equal to the preset clarity threshold; and a picture recording subunit, used to directly record the portrait picture as the clear portrait picture if the clarity of the portrait picture is greater than or equal to the preset clarity threshold.

[0127] Preferably, the personnel registration device further comprises: a sample image acquisition module for acquiring a sample image set; the sample image set comprises at least one sample blurred image and a sample clear image corresponding to the sample blurred image; a model acquisition module for acquiring a preset recognition model including initial parameters; the preset recognition model comprises a first image conversion network and a second image conversion network; an image conversion module for inputting the sample blurred image into the first image conversion network to obtain a predicted blurred image corresponding to the sample blurred image; inputting the sample clear image into the second image conversion network to obtain a predicted clear image corresponding to the sample clear image; loss A parameter determination module is used to determine the first loss parameter of the first image conversion network based on the sample blurred image and the predicted blurred image; and to determine the second loss parameter of the second image conversion network based on the sample clear image and the predicted clear image; a predicted loss value determination module is used to determine the predicted loss value of the preset recognition model based on the first loss parameter and the second loss parameter; and a parameter updating module is used to iteratively update the initial parameters in the preset recognition model when the predicted loss value does not reach the preset convergence condition, until the predicted loss value reaches the convergence condition, and then record the preset recognition model after convergence as the preset image conversion model.

[0128] Preferably, the feature comparison unit includes: a portrait recognition subunit, which is used to perform portrait recognition on the clear portrait picture to obtain at least one captured portrait picture; the captured portrait picture refers to a portrait picture captured from the clear portrait picture and containing only one photographed subject; an image grouping subunit, which is used to associate and record the captured portrait pictures containing the same photographed subject as a photographed portrait group; a feature comparison subunit, which is used to determine, for each of the photographed portrait groups, the image similarity between the person registration picture and the captured portrait picture in the photographed portrait group, and determine the image comparison result between the person registration picture and the clear portrait picture based on the image similarity.

[0129] Preferably, the image comparison result is associated with the shooting time of the portrait picture; the registration status confirmation module 40 includes: obtaining a preset time range, and determining at least one historical shooting time according to the shooting time of the portrait picture and the preset time range; the historical shooting time is earlier than the shooting time of the portrait picture; a picture acquisition unit, used to obtain a historical picture corresponding to the historical shooting time; the historical picture contains at least one shooting object; a portrait tracking unit, used to perform portrait tracking on the shooting object contained in the historical picture according to the historical picture and the portrait picture, and obtain a portrait tracking result corresponding to the shooting object contained in the historical picture; a registration status confirmation unit, used to determine the community registration status of the shooting object contained in the portrait picture in the target community according to the portrait tracking result, the image comparison result and the third-party registration information.

[0130] Preferably, the community registration status includes: unregistered status, registered status, unregistered status and cancelled status, and the third-party registration information includes a cancelled personnel information table and an existing personnel information table; the registration status confirmation unit includes: an identity registration confirmation subunit, for determining the identity registration result of the photographed object contained in the portrait photograph according to the portrait photograph, the cancelled personnel information table and the existing personnel information table; an unregistered status confirmation subunit, for determining that the community registration status is an unregistered status when the portrait tracking result indicates a tracking failure, the image comparison result indicates a comparison success, and the identity registration result indicates that the portrait photograph successfully matches the existing personnel information table but fails to match the cancelled personnel information table; a cancelled status confirmation subunit, for determining that the community registration status is an unregistered status when the portrait tracking result indicates a tracking failure, the image comparison result indicates a comparison success, and the identity registration result indicates that the portrait photograph successfully matches the existing personnel information table but fails to match the cancelled personnel information table; The community registration status is determined to be the cancelled status when the characterization comparison is successful, and the identity registration result characterizes that the portrait picture fails to match the existing personnel information table but successfully matches the cancelled personnel information table; the unregistered status confirmation subunit is used to determine that the community registration status is the unregistered status when the portrait tracking result characterizes successful tracking, the image comparison result characterizes successful comparison, and the identity registration result characterizes that the personnel picture fails to match both the existing personnel information table and the cancelled personnel information table; the registered status confirmation subunit is used to determine that the community registration status is the registered status when the portrait tracking result characterizes successful tracking, the image comparison result characterizes successful comparison, and the identity registration result characterizes that the personnel picture successfully matches the existing personnel information table but fails to match the cancelled personnel information table.

[0131] The specific definition of the personnel registration device can be found in the definition of the personnel registration method above and will not be repeated here. Each module in the above-mentioned personnel registration device can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor of the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each of the above modules.

[0132] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data used in the personnel registration method in the above embodiment. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a personnel registration method is implemented.

[0133] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the personnel registration method in the above embodiment is implemented.

[0134] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the personnel registration method in the above embodiment is implemented.

[0135] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0136] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0137] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A personnel registration method, characterized in that: include: Obtain identity registration information for target communities; The identity registration information includes a registration picture of at least one community member belonging to the target community; Obtaining a portrait picture taken within a preset geographical range of the target community; the portrait picture includes at least one photographed subject; Performing feature comparison between the personnel registration picture and the portrait picture to obtain an image comparison result between the personnel registration picture and the portrait picture; Obtaining third-party registration information corresponding to the target community, and determining the community registration status of the subject included in the portrait image in the target community based on the image comparison result and the third-party registration information; The preset third-party registration information is the registration information in the public security system of the target community. The preset third-party registration information is different from the identity registration information of the target community; The image comparison result is associated with the shooting time of the portrait picture; The determining, based on the image comparison result and the third-party registration information, the community registration status of the subject included in the portrait image in the target community includes: Obtaining a preset time range, and determining at least one historical shooting time according to the shooting time of the portrait picture and the preset time range; the historical shooting time is earlier than the shooting time of the portrait picture; Acquire a historical picture corresponding to the historical shooting time; the historical picture includes at least one shooting object; performing portrait tracking on a subject included in the historical pictures according to the historical pictures and the portrait pictures, and obtaining a portrait tracking result corresponding to the subject included in the historical pictures; the portrait tracking result represents the number of times the subject appears in the historical pictures and the portrait pictures; The community registration status of the subject included in the portrait image in the target community is determined based on the portrait tracking result, the image comparison result, and the third-party registration information.

2. The personnel registration method according to claim 1, wherein: The performing feature comparison between the personnel registration picture and the portrait picture to obtain an image comparison result between the personnel registration picture and the portrait picture includes: Obtaining the clarity of the portrait picture, and generating a portrait clarity picture that meets a preset clarity condition based on the clarity of the portrait picture; A feature comparison is performed between the clear portrait picture and the personnel registration picture to obtain an image comparison result between the personnel registration picture and the clear portrait picture.

3. The personnel registration method according to claim 2, wherein: The generating of a clear portrait picture meeting a preset clarity condition based on the clarity of the portrait picture includes: Comparing the clarity of the portrait image with a preset clarity threshold; If the clarity of the portrait picture is less than the preset clarity threshold, input the portrait picture into a preset image conversion model to generate a portrait clear picture corresponding to the portrait picture through the preset image conversion model; the clarity of the portrait clear picture is greater than or equal to the preset clarity threshold; If the clarity of the portrait picture is greater than or equal to the preset clarity threshold, the portrait picture is directly recorded as the clear portrait picture.

4. The personnel registration method according to claim 3, wherein: Before inputting the portrait picture into the preset image conversion model, the method further includes: Acquire a sample image set; the sample image set includes at least one sample blurred image and a sample clear image corresponding to the sample blurred image; Obtaining a preset recognition model including initial parameters; the preset recognition model includes a first image conversion network and a second image conversion network; Inputting the sample blurred image into the first image conversion network to obtain a predicted blurred image corresponding to the sample blurred image; inputting the sample clear image into the second image conversion network to obtain a predicted clear image corresponding to the sample clear image; Determine a first loss parameter of the first image conversion network according to the sample blurred image and the predicted blurred image; determine a second loss parameter of the second image conversion network according to the sample clear image and the predicted clear image; Determining a predicted loss value of the preset recognition model according to the first loss parameter and the second loss parameter; When the predicted loss value does not reach the preset convergence condition, the initial parameters in the preset recognition model are iteratively updated until the predicted loss value reaches the convergence condition, and the preset recognition model after convergence is recorded as the preset image conversion model.

5. The personnel registration method according to claim 2, wherein: The performing feature comparison between the clear portrait picture and the personnel registration picture to obtain an image comparison result between the personnel registration picture and the clear portrait picture includes: Performing portrait recognition on the clear portrait picture to obtain at least one captured portrait picture; the captured portrait picture is a portrait picture captured from the clear portrait picture and containing only one subject; Associate and record the captured portrait images of the same subject as a portrait group; For each of the photographed portrait groups, the image similarity between the person registration picture and the captured portrait pictures in the photographed portrait group is determined, and the image comparison result between the person registration picture and the clear portrait picture is determined based on the image similarity.

6. The personnel registration method according to claim 1, wherein: The community registration status includes: unregistered status, registered status, unregistered status and cancelled status, and the third-party registration information includes a cancelled personnel information table and an existing personnel information table; The determining, based on the portrait tracking result, the image comparison result, and the third-party registration information, of the community registration status of the subject included in the portrait image in the target community includes: Determining the identity registration result of the subject included in the portrait picture according to the portrait picture, the deregistered person information table, and the existing person information table; When the portrait tracking result indicates that tracking has failed, the image comparison result indicates that the comparison has been successful, and the identity registration result indicates that the portrait image has successfully matched the existing personnel information table but failed to match the deregistered personnel information table, the community registration status is determined to be a non-deregistered status; When the portrait tracking result indicates that tracking has failed, the image comparison result indicates that comparison has been successful, and the identity registration result indicates that the portrait image has failed to match the existing personnel information table but has successfully matched the deregistered personnel information table, the community registration status is determined to be the deregistered status; When the portrait tracking result indicates that the tracking is successful, the image comparison result indicates that the comparison is successful, and the identity registration result indicates that the photographed image of the person fails to match both the existing person information table and the deregistered person information table, determining that the community registration status is the unregistered status; When the portrait tracking result indicates successful tracking, the image comparison result indicates successful comparison, and the identity registration result indicates that the photograph of the person successfully matches the existing person information table but fails to match the deregistered person information table, the community registration status is determined to be the registered status.

7. A personnel registration device, characterized in that: include: A registration information acquisition module is used to acquire identity registration information of a target community; the identity registration information includes a registration photo of at least one community member belonging to the target community; An image acquisition module is configured to acquire a portrait image taken within a preset geographical range of the target community; the portrait image includes at least one photographed subject; An image feature comparison module is used to perform feature comparison between the personnel registration picture and the portrait picture to obtain an image comparison result between the personnel registration picture and the portrait picture; a registration status confirmation module, configured to obtain third-party registration information corresponding to the target community, and determine the community registration status corresponding to each of the photographed subjects based on the image comparison result and the third-party registration information; The preset third-party registration information is the registration information in the public security system of the target community. The preset third-party registration information is different from the identity registration information of the target community; The image comparison result is associated with the shooting time of the portrait picture; The determining, based on the image comparison result and the third-party registration information, the community registration status of the subject included in the portrait image in the target community includes: Obtaining a preset time range, and determining at least one historical shooting time according to the shooting time of the portrait picture and the preset time range; the historical shooting time is earlier than the shooting time of the portrait picture; Acquire a historical picture corresponding to the historical shooting time; the historical picture includes at least one shooting object; performing portrait tracking on a subject included in the historical pictures according to the historical pictures and the portrait pictures, and obtaining a portrait tracking result corresponding to the subject included in the historical pictures; the portrait tracking result represents the number of times the subject appears in the historical pictures and the portrait pictures; The community registration status of the subject included in the portrait image in the target community is determined based on the portrait tracking result, the image comparison result, and the third-party registration information.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the personnel registration method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the personnel registration method according to any one of claims 1 to 6 is implemented.

Images