Portrait clustering method and apparatus, electronic device, and storage medium

Abstract

A portrait clustering method and device, electronic equipment and storage medium are provided to improve the recall rate of portrait clustering. In the present application, first portrait information collected by a first camera in a first time period is obtained, and second portrait information collected by a second camera in the first time period is obtained. According to the correlation degree of the first camera and the second camera and the similarity comparison result of the first portrait information and the second portrait information, the target similarity of the first portrait information and the second portrait information is determined. According to the target similarity, the first portrait information and the second portrait information are clustered. Since the correlation of the cameras is combined for clustering, the recall rate of portrait clustering can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image clustering, and in particular to a portrait clustering method and device, an electronic device, and a storage medium. BACKGROUND

[0002] With the joint promotion of image recognition and clustering technology, real-time computing technology, and hardware storage media, large-scale portrait clustering has gradually become possible. At present, an information library with one file per person is usually established, and images of the same person in the information library belong to the same file. In order to make the file real-time, image clustering is usually performed when a real-time picture enters the image library, that is, the pictures of one person are clustered into one cluster. Then the cluster is compared with all portrait files in the historical database, and the real-time picture cluster that matches the historical file is merged into the corresponding historical file, and the real-time picture cluster that does not match the historical file is established as a new file.

[0003] Traditional portrait clustering technology focuses on comparing image feature values from images themselves to realize portrait clustering. In the clustering process, the picture records of the same person may be divided into multiple files (usually described by the recall rate). The recall rate is an important indicator for evaluating the quality of file clustering. How to further improve the recall rate has become a problem to be solved. SUMMARY

[0004] The purpose of the present application is to provide a portrait clustering method, device, electronic device, and storage medium for improving the recall rate of portrait clustering.

[0005] In a first aspect, a portrait clustering method is provided, which can include: acquiring first portrait information collected by a first camera in a first time period, and acquiring second portrait information collected by a second camera in the first time period. According to the correlation degree of the first camera and the second camera and the similarity comparison result of the first portrait information and the second portrait information, a target similarity of the first portrait information and the second portrait information is determined. The correlation degree is determined based on information of people passing through the first camera and the second camera in a second time period.

[0006] Compared with the scheme of clustering portrait information only by the similarity comparison result of the portrait information, the present application can determine the target similarity of the first portrait information and the second portrait information based on the correlation of the first camera and the second camera. For example, when the correlation degree of the first camera and the second camera is high, the similarity of the two portrait information of the two cameras in the first time period can be stretched accordingly, so as to improve the recall rate of portrait clustering and more accurately cluster the portrait information.

[0007] To illustrate the above beneficial effects, the following is used as an example: the similarity between the first portrait information and the second portrait information is 88 points, and if the similarity is not stretched, the first portrait information and the second portrait information will be classified into two archives.

[0008] However, if the correlation degree of the first card hole and the second card hole is high, the similarity of the two portrait information can be stretched, for example, the similarity of the two portrait information can be stretched from 88 points to 91 points. If the similarity of the two portrait information is 90 points, the two portrait information can be determined as portrait information of the same person. It can be seen that since 91 points is greater than 90 points, the two portrait information can be determined as portrait information of the same person, and thus belong to the same archive.

[0009] As can be seen from the above example, in the embodiment of the present application, the similarity of two graphics is stretched based on the correlation of the card hole, which can reduce the phenomenon of classifying multiple portrait information of one person into multiple archives, thereby improving the recall rate. On the other hand, when the correlation degree of the two card holes is high, the probability that the two relatively similar portrait information of the two card holes belong to the same person is also relatively high. Based on this, the portrait information can be more accurately clustered in the embodiment of the present application.

[0010] In a possible implementation, the correlation degree of the first card hole and the second card hole is used to indicate the proportion of people passing through the second card hole among people passing through the first card hole in the second time period, and / or the proportion of people passing through the first card hole among people passing through the second card hole in the second time period. According to the target similarity, the first portrait information and the second portrait information are clustered.

[0011] In a possible implementation, the correlation degree of the first card hole and the second card hole is determined according to at least one of the following: a first ratio of a first quantity to a second quantity, the first quantity being the number of people passing through the first card hole and passing through the second card hole in the second time period, and the second quantity being the number of people passing through the first card hole in the second time period; a second ratio of the first quantity to a third quantity, the third quantity being the number of people passing through the second card hole in the second time period; or information indicating the average of the first ratio and the second ratio.

[0012] The first ratio can indicate how many of the people passing through the first portal also pass through the second portal in the second time period. The second ratio can indicate how many of the people passing through the second portal also pass through the first portal in the second time period. Therefore, the degree of correlation between the first portal and the second portal can be determined more accurately by using at least one of the above. Further, the second time period can be relatively short, such as one hour, two hours, etc., and the correlation between the first portal and the second portal can be inferred more accurately by data, further reducing errors.

[0013] In a possible implementation, the information indicating the average of the first ratio and the second ratio includes at least one of the following: the average of the first ratio and the second ratio; or, the harmonic mean of the first ratio and the second ratio.

[0014] In this way, errors in determining the degree of correlation between the first portal and the second portal can be reduced. In another aspect, for example, when the first portal is the portal of the entrance of a hotel and the second portal is the portal of the corridor of the hotel, the first portal collects more portrait information. If the harmonic mean of the first ratio and the second ratio is used to determine the degree of correlation between the first portal and the second portal, the harmonic mean of the first ratio and the second ratio is more biased towards a smaller value, and therefore the error in determining the degree of correlation between the first portal and the second portal can be further reduced.

[0015] In a possible implementation, according to the degree of correlation between the first portal and the second portal and the similarity comparison result of the first portrait information and the second portrait information, a target similarity of the first portrait information and the second portrait information is determined, including: performing a similarity comparison on the first portrait information and the second portrait information to obtain a first similarity. In a case where the degree of correlation between the first portal and the second portal indicates a correlation greater than a preset correlation threshold, a second similarity corresponding to the first similarity is determined as the target similarity according to a similarity mapping relationship corresponding to the degree of correlation between the first portal and the second portal, wherein the similarity mapping relationship includes a corresponding relationship between the first similarity and the second similarity. In a possible implementation, the similarity indicated by the second similarity is higher than the similarity indicated by the first similarity.

[0016] In a case where the degree of correlation between the first portal and the second portal indicates a correlation greater than a preset correlation threshold, it indicates that the first portal and the second portal have a certain correlation. In this case, the similarity of the two portrait information of the two portals in the first time period can be stretched accordingly, thereby improving the recall rate.

[0017] In a possible implementation, after the similarity comparison of the first portrait information and the second portrait information is performed to obtain a first similarity, the method further includes: in a case where the correlation indicated by the correlation degree of the first passageway and the second passageway is not greater than a preset correlation threshold, determining the first similarity as the target similarity.

[0018] When the correlation indicated by the correlation degree of the first passageway and the second passageway is not greater than the preset correlation threshold, it indicates that the first passageway and the second passageway are weakly correlated, and in this case, the probability that the two pieces of portrait information of the two passageways belong to the same person is relatively low. In this case, clustering can be performed only based on the similarity comparison result of the first portrait information and the second portrait information, and the similarity of the portrait information is no longer stretched, so that the clustering of the portrait information can be more accurate.

[0019] In a second aspect, a device for portrait clustering is provided, which includes an acquisition module and a clustering module. The acquisition module is configured to acquire first portrait information collected by a first passageway in a first time period, and acquire second portrait information collected by a second passageway in the first time period. The clustering module is configured to determine a target similarity of the first portrait information and the second portrait information according to a correlation degree of the first passageway and the second passageway, and a similarity comparison result of the first portrait information and the second portrait information. The correlation degree is determined based on information of a person passing through the first passageway and a person passing through the second passageway in a second time period.

[0020] Compared with a scheme of clustering portrait information only based on a similarity comparison result of the portrait information, the present application can determine the target similarity of the first portrait information and the second portrait information based on the correlation of the first passageway and the second passageway, so that the recall rate of portrait clustering can be improved.

[0021] In a possible implementation, the correlation degree of the first passageway and the second passageway is used to indicate a proportion of a person passing through the second passageway in persons passing through the first passageway in the second time period, and / or a proportion of a person passing through the first passageway in persons passing through the second passageway in the second time period; and the first portrait information and the second portrait information are clustered according to the target similarity.

[0022] In a possible implementation, the correlation degree of the first passageway and the second passageway is determined according to at least one of the following: a first ratio of a first quantity to a second quantity, the first quantity being a quantity of persons passing through the first passageway and passing through the second passageway in the second time period, and the second quantity being a quantity of persons passing through the first passageway in the second time period; a second ratio of the first quantity to a third quantity, the third quantity being a quantity of persons passing through the second passageway in the second time period; or information indicating an average of the first ratio and the second ratio.

[0023] The first ratio can indicate how many of the people passing through the first portal also pass through the second portal in the second time period. The second ratio can indicate how many of the people passing through the second portal also pass through the first portal in the second time period. Therefore, the degree of correlation between the first portal and the second portal can be determined more accurately by using at least one of the above. Further, the second time period can be relatively short, such as one hour, two hours, etc., and the correlation between the first portal and the second portal can be inferred more accurately by data, further reducing errors.

[0024] In a possible implementation, the information indicating the average of the first ratio and the second ratio includes at least one of: the average of the first ratio and the second ratio; or the harmonic mean of the first ratio and the second ratio.

[0025] In this way, errors in determining the degree of correlation between the first portal and the second portal can be reduced. In another aspect, for example, when the first portal is a portal of a hotel entrance and the second portal is a portal of a hotel corridor, the first portal collects more portrait information. If the harmonic mean of the first ratio and the second ratio is used to determine the degree of correlation between the first portal and the second portal, the harmonic mean of the first ratio and the second ratio is more biased towards a smaller value, and therefore the error in determining the degree of correlation between the first portal and the second portal can be further reduced.

[0026] In a possible implementation, the clustering module is specifically configured to: perform a similarity comparison on the first portrait information and the second portrait information to obtain a first similarity. In a case where the degree of correlation between the first portal and the second portal indicates a correlation greater than a preset correlation threshold, a second similarity corresponding to the first similarity is determined as a target similarity according to a preset similarity mapping relationship. The similarity mapping relationship includes a corresponding relationship between the first similarity and the second similarity. In a possible implementation, the second similarity indicates a higher degree of similarity than the first similarity.

[0027] In a case where the degree of correlation between the first portal and the second portal indicates a correlation greater than a preset correlation threshold, it indicates that the first portal and the second portal have a certain correlation. In this case, the similarity of the two portrait information of the two portals in the first time period can be stretched accordingly, thereby improving the recall rate.

[0028] In a possible implementation, the clustering module is further configured to: in a case where the degree of correlation between the first portal and the second portal indicates a correlation not greater than a preset correlation threshold, determine the first similarity as the target similarity.

[0029] When the degree of correlation of the first and second card slots indicates that the correlation is not greater than a preset correlation threshold, it is determined that the first and second card slots are weakly correlated, and the two pieces of portrait information of the two card slots belong to the same person with a lower probability. In this case, clustering can be performed only based on the similarity comparison result of the first and second pieces of portrait information, and the similarity of the portrait information is no longer stretched, so that the clustering of the portrait information can be more accurate.

[0030] In a third aspect, an electronic device is provided, and the electronic device includes:

[0031] a memory configured to store program instructions;

[0032] a processor configured to invoke the program instructions stored in the memory and perform the steps included in the method of any one of the first aspect according to the obtained program instructions.

[0033] In a fourth aspect, a computer readable storage medium is provided, and the computer readable storage medium stores computer executable instructions for causing a computer to perform the steps included in the method of any one of the first aspect.

[0034] In a fifth aspect, a computer program product including instructions is provided, which, when executed on a computer, causes the computer to perform the identity archiving method described in the various possible implementation manners. BRIEF DESCRIPTION OF DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application.

[0036] Figure 1 An architecture schematic diagram of a person portrait clustering applicable scenario provided by an embodiment of the present application;

[0037] Figure 2 A possible flowchart of a person portrait clustering method provided by an embodiment of the present application;

[0038] Figure 3 A possible flowchart of a person portrait clustering method provided by an embodiment of the present application;

[0039] Figure 4 A possible flowchart of a person portrait clustering method provided by an embodiment of the present application;

[0040] Figure 5 A possible flowchart of a person portrait clustering method provided by an embodiment of the present application;

[0041] Figure 6A possible structural schematic diagram of a device for portrait clustering provided by an embodiment of the present application;

[0042] Figure 7 A possible structural schematic diagram of a device for portrait clustering in another embodiment of the present application. DETAILED DESCRIPTION

[0043] In order for those skilled in the art to better understand the technical solutions of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings.

[0044] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Rather, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0045] For ease of understanding, the professional terms in the present application are first explained:

[0046] Portrait information generally includes face image information and body image information. Portrait information can also be referred to as portrait data. Portrait information can include portrait images. Face image information can also be referred to as face image data, and face image information can also include face images. Body image information can also be referred to as body image data, and body image information can also include body images. The first portrait information and the second portrait information mentioned in the embodiments of the present application are two portrait information, wherein "first" and "second" are merely used to distinguish the first portrait information and the second portrait information.

[0047] Clustering portrait information: a process of comparing and grouping image data in a library to form a plurality of personnel image sets.

[0048] Portfolio: a set produced by face clustering, identifying a set of virtual people.

[0049] The inventors have found that in order to improve the accuracy and recall rate of clustering portfolios, the following solutions can be used:

[0050] Images archived to the archive are categorized according to angle, attributes, etc., with a maximum of two high-quality images retained for each category. When a higher-quality archived image appears, it is updated and replaced. If there are not enough images in a category and the upper limit is not reached, new images are added. The images in these archives are fused to generate the archive features of this archive, thereby improving the accuracy and recall rate of the archive aggregation.

[0051] The above approach has drawbacks. For example, it lacks a good model for determining whether an image is of higher quality. For instance, how do we classify an image as better when both portraits are in profile? Furthermore, when both portraits are in profile, the fewer facial features depicted in profile views result in less facial information in the feature set, thus failing to fully represent the corresponding individuals. This limits the positive impact on clustering and makes it easier to introduce more errors.

[0052] In view of this, this application proposes a facial image clustering method, apparatus, electronic device and storage medium for clustering facial image information based on the correlation between checkpoints, thereby improving the recall rate.

[0053] The following is a brief introduction to the application scenarios to which the technical solutions of the embodiments of this application are applicable. It should be noted that the application scenarios described below are only for illustrating the embodiments of this application and are not intended to limit the scope. In specific implementation, the technical solutions provided by the embodiments of this application can be flexibly applied according to actual needs.

[0054] like Figure 1 The diagram shown illustrates an application scenario of the facial clustering method in this embodiment. The diagram includes: a server 101, a storage unit 102, and a checkpoint. Figure 1 The diagram shows two checkpoints, namely the first checkpoint 103 and the second checkpoint 104. In practical applications, this scenario may include more checkpoints, and this embodiment does not impose any limitations.

[0055] A checkpoint can also be called a checkpoint device or an image acquisition device. Checkpoints can be, for example, surveillance equipment or facial recognition cameras, and other devices used for image acquisition.

[0056] Server 101 and the checkpoint can transmit data over a network. The server can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud services, cloud databases, cloud computing, etc.

[0057] The description in this application is only detailed for a single server, a first camera and a second camera, but those skilled in the art should understand that the camera, the server 101 and the memory 102 shown are intended to represent the operation of the camera, the server and the memory involved in the technical solution of the present application. The single server and the memory are detailed at least for the convenience of illustration, and not to imply a limitation on the number, type or location of the camera and the server. It should be noted that if additional modules are added to the illustrated environment or individual modules are removed therefrom, the underlying concept of the example embodiments of the present application will not change.

[0058] It should be noted that the memory in the embodiments of the present application may, for example, be a cache system, a hard disk storage, a memory storage, etc. In addition, the portrait clustering method proposed in the present application is not only applicable to the application scenarios shown, but also applicable to any device with portrait clustering needs. Figure 1

[0059] The portrait clustering method provided by the embodiments of the present application will be introduced below in conjunction with the accompanying drawings of the specification. The portrait clustering method provided by the embodiments of the present application can be executed by the server 101 in the server 101, or can be executed by a module, a chip or a unit in the server 101. Please refer to FIG. 1, the flow of the portrait clustering method in the embodiments of the present application is described as follows: Figure 1 Figure 2

[0060] S201, the server acquires first portrait information collected by a first camera in a first time period.

[0061] S201, the server acquires second portrait information collected by a second camera in the first time period.

[0062] S203, the server determines a target similarity of the first portrait information and the second portrait information according to a correlation degree of the first camera and the second camera, and a similarity comparison result of the first portrait information and the second portrait information.

[0063] The correlation degree can be determined based on information of people passing through the first camera and people passing through the second camera in a second time period. In a possible implementation, the correlation degree of the first camera and the second camera is used to indicate a proportion of people passing through the second camera among people passing through the first camera in the second time period, and / or a proportion of people passing through the first camera among people passing through the second camera in the second time period.

[0064] S204, the server clusters the first portrait information and the second portrait information according to the target similarity.

[0065] ​​​In S204, in a case where the target similarity is greater than the preset similarity threshold, it can be determined that the first portrait information and the second portrait information are portrait information of the same person. The first portrait information and the second portrait information can be determined as portrait information of the same person, and then can be classified into the same archive.

[0066] Compared with a scheme of clustering portrait information only through a similarity comparison result of portrait information, the application can determine the target similarity of the first portrait information and the second portrait information based on the correlation of the first lens and the second lens, so that the recall rate of portrait clustering can be improved.

[0067] Before S203, based on the above content, in the embodiment of the application, the correlation between two lenses can be calculated according to historical data. In actual application, multiple lenses are involved, and in the embodiment of the application, a possible scheme for calculating the correlation between two lenses is demonstrated by taking the correlation between the first lens and the second lens as an example. The first lens can be the first lens 103 in Figure 1 , and the second lens can be the second lens 104 in Figure 1 . Figure 3 An exemplary method flow diagram for calculating the correlation between the first lens and the second lens is shown. The portrait clustering method provided by the embodiment of the application can be executed by the server 101 in Figure 1 , or can be executed by a module, a chip or a unit in the server 101. Please refer to Figure 3 :

[0068] S301, the server counts a first number of people passing through the first lens and passing through the second lens in a second time period.

[0069] In order to distinguish, in the embodiment of the application, the number of people passing through the first lens and passing through the second lens in the second time period is referred to as the first number.

[0070] In the embodiment of the application, the second time period can be a historical time period, and the second time period can be one hour, two hours or half an hour, etc. Further, if the length of the second time period is set to be relatively short, such as one hour or two hours, the correlation between the first lens and the second lens can be inferred more accurately through data, and the error can be further reduced.

[0071] S302, the server counts a second number of people passing through the first lens in the second time period.

[0072] In order to distinguish, in the embodiment of the application, the number of people passing through the first lens in the second time period is referred to as the second number. The second number is not less than the first number.

[0073] S303, the server counts a third quantity of people passing through the second passageway in the second time period.

[0074] In the embodiments of the present application, the quantity of people passing through the second passageway in the second time period is referred to as the third quantity for the sake of distinction. The third quantity is not less than the first quantity. There is no certain size relationship between the third quantity and the second quantity, and the third quantity can be greater than, equal to, or less than the second quantity.

[0075] S304, the server calculates a first ratio of the first quantity to the second quantity.

[0076] In the embodiments of the present application, the ratio of the first quantity to the second quantity is referred to as the first ratio for the sake of distinction. The first ratio can indicate how many people passing through the first passageway pass through the second passageway in the second time period.

[0077] S305, the server calculates a second ratio of the first quantity to the third quantity.

[0078] In the embodiments of the present application, the ratio of the first quantity to the third quantity is referred to as the second ratio for the sake of distinction. The second ratio can indicate how many people passing through the second passageway pass through the first passageway in the second time period.

[0079] S306, the server determines the degree of correlation of the first passageway and the second passageway according to at least one of the first ratio or the second ratio.

[0080] In S306, the degree of correlation of the first passageway and the second passageway is determined according to at least one of the following:

[0081] the first ratio;

[0082] the second ratio; or

[0083] information indicating the average of the first ratio and the second ratio.

[0084] The information indicating the average of the first ratio and the second ratio includes at least one of the average of the first ratio and the second ratio, or the harmonic mean of the first ratio and the second ratio. In this way, the error in the process of determining the degree of correlation of the first passageway and the second passageway can be reduced. On the other hand, for example, when the first passageway is the passageway of the entrance of a hotel, and the second passageway is the passageway of a corridor of the hotel, the first passageway collects more image information. If the harmonic mean of the first ratio and the second ratio is used to determine the degree of correlation of the first passageway and the second passageway, the harmonic mean of the first ratio and the second ratio is more biased towards a smaller value, and thus the error in the process of determining the degree of correlation of the first passageway and the second passageway can be further reduced.

[0085] For example, a correlation threshold can be set, such as a first correlation threshold. The correlation of the first and second card holes can be the harmonic mean of the first and second ratios, in which case, when the harmonic mean of the first and second ratios is greater than the preset first correlation threshold, it can be determined that the correlation degree of the first and second card holes is strong correlation, otherwise, it is determined that the correlation degree of the first and second card holes is weak correlation.

[0086] For another example, the correlation of the first and second card holes can be the average of the first and second ratios, in which case, when the average of the first and second ratios is greater than the preset second correlation threshold, it can be determined that the correlation degree of the first and second card holes is strong correlation, otherwise, it is determined that the correlation degree of the first and second card holes is weak correlation.

[0087] For another example, the correlation of the first and second card holes can be the first ratio or the second ratio, taking the first ratio as an example, in which case, when the first ratio is greater than the preset third correlation threshold, it can be determined that the correlation degree of the first and second card holes is strong correlation, otherwise, it is determined that the correlation degree of the first and second card holes is weak correlation.

[0088] The first, second and third correlation thresholds described above are only used to distinguish the three correlation thresholds, any two of the three correlation thresholds can be the same or different, and the embodiments of the present application do not limit.

[0089] The above examples only show several examples of determining the correlation degree, and in actual applications, the correlation of the first and second card holes can also be calculated according to other formulas and operation rules, such as setting weights for the first and second ratios, weighting and adding the first and second ratios to obtain the correlation of the first and second card holes, and further comparing the correlation of the first and second card holes with the preset correlation threshold to determine the correlation degree between the first and second card holes.

[0090] The above is only an example, and in actual applications, multiple correlation thresholds can also be set, such as a fourth correlation threshold and a fifth correlation threshold. The correlation of the first and second sockets can be the harmonic mean of the first ratio and the second ratio. In this case, when the harmonic mean of the first ratio and the second ratio is greater than the fourth correlation threshold, it can be determined that the correlation degree of the first and second sockets is strong correlation. When the harmonic mean of the first ratio and the second ratio is not greater than the fourth correlation threshold and is greater than the fifth correlation threshold, it can be determined that the correlation degree of the first and second sockets is moderate correlation. When the harmonic mean of the first ratio and the second ratio is not greater than the fifth correlation threshold, it can be determined that the correlation degree of the first and second sockets is weak correlation. In this way, more levels of correlation degrees can be divided to improve the accuracy of the calculation.

[0091] For S203 described above, Figure 4 An example shows a possible method flow diagram provided by the embodiment of the application for determining the similarity of two portrait information based on the correlation degree of the sockets. The method can be executed by the server 101 in Figure 1 The server 101 in the embodiment of the application can also be executed by a module, a chip or a unit in the server 101. Figure 4 The first portrait information collected by the first socket and the second portrait information collected by the second socket in S201 are taken as an example for illustration, and the determination scheme of the similarity between the two portrait information collected by the other two sockets is similar, which will not be repeated. Next, see Figure 4 The method comprises the following steps:

[0092] S401, the server compares the similarity of the first portrait information and the second portrait information, and obtains a first similarity.

[0093] In S401, the similarity of the two portrait information can be determined by various schemes, such as cosine similarity (in this case, the first similarity can be the cosine similarity value of the first portrait information and the second portrait information), Manhattan distance, or inner product similarity, etc. The embodiment of the application does not make any limitation.

[0094] S402, the server determines whether the correlation indicated by the correlation degree of the first socket and the second socket is greater than a preset correlation threshold.

[0095] If yes, S403 is executed;

[0096] If no, S404 is executed.

[0097] S403, the server determines the second similarity corresponding to the first similarity as a target similarity according to the similarity mapping relationship corresponding to the correlation degree of the first socket and the second socket. The similarity mapping relationship includes the corresponding relationship between the first similarity and the second similarity.

[0098] The similarity degree indicated by the second similarity indication is higher than the similarity degree indicated by the first similarity indication.

[0099] In the embodiments of the present application, a correlation threshold value can be preset. When the correlation of the first card hole and the second card hole is greater than the correlation threshold value, it can be determined that the first card hole and the second card hole are strongly correlated, otherwise, they are weakly correlated. In this case, in S402, when it is determined that the correlation degree indicated by the correlation of the first card hole and the second card hole is greater than the preset correlation threshold value, it can also be replaced by: the correlation degree of the first card hole and the second card hole is strongly correlated.

[0100] In another possible implementation, a plurality of correlation threshold values can be preset in the embodiments of the present application. In this case, the correlation of the first card hole and the second card hole greater than the correlation threshold value in S402 can include that the correlation degree of the first card hole and the second card hole is not weakly correlated.

[0101] For example, in the embodiments of the present application, a fourth correlation threshold value and a fifth correlation threshold value are set respectively, the correlation of the first card hole and the second card hole can be the harmonic mean of the first ratio and the second ratio. In this case, when the harmonic mean of the first ratio and the second ratio is greater than the fourth correlation threshold value, it can be determined that the correlation degree of the first card hole and the second card hole is strongly correlated. When the harmonic mean of the first ratio and the second ratio is not greater than the fourth correlation threshold value but greater than the fifth correlation threshold value, it can be determined that the correlation degree of the first card hole and the second card hole is moderately correlated. When the harmonic mean of the first ratio and the second ratio is not greater than the fifth correlation threshold value, it can be determined that the correlation degree of the first card hole and the second card hole is weakly correlated. In a possible implementation, the correlation threshold value in S402 can be the fifth correlation threshold value. In another possible implementation, the correlation threshold value in S402 can be the fourth correlation threshold value.

[0102] In the embodiments of the present application, one or more similarity mapping relationships can be preset. The similarity mapping relationship includes the corresponding relationship of two similarities. For example, in the embodiments of the present application, the corresponding similarity mapping relationship can be set for strong correlation, and the corresponding similarity mapping relationship can also be set for moderate correlation. In this case, when the correlation degree indicated by the correlation of the first card hole and the second card hole is greater than the fourth threshold value, that is, the first card hole and the second card hole are strongly correlated, the corresponding similarity mapping relationship can be determined according to the strong correlation degree to determine the second similarity corresponding to the first similarity. When the correlation degree indicated by the correlation of the first card hole and the second card hole is not greater than the fourth threshold value but greater than the third threshold value, that is, the first card hole and the second card hole are moderately correlated, the corresponding similarity mapping relationship can be determined according to the moderate correlation to determine the second similarity corresponding to the first similarity.

[0103] In the embodiments of the present application, the stronger the correlation between the two card slots is, the higher the similarity can be stretched to, that is, the stronger the correlation between the two card slots is, the more likely the two pieces of portrait information passing through the two card slots belong to the same person.

[0104] For example, the strong correlation corresponds to the mapping relationship of the third similarity and the fourth similarity, and the moderate correlation corresponds to the mapping relationship of the fifth similarity and the sixth similarity. The fourth similarity is greater than the third similarity, and the sixth similarity is greater than the fifth similarity. The difference between the fourth similarity and the third similarity can be greater than the difference between the sixth similarity and the fifth similarity. For example, the third similarity is 88 points, and the fourth similarity can be pulled up by 3 points to 91 points. The fifth similarity is 85 points, and the sixth similarity can be pulled up by 2 points to 87 points. That is, when the two card slots are strongly correlated, the similarity between the two pieces of portrait information under the two card slots can be stretched by 3 points. When the two card slots are moderately correlated, the similarity between the two pieces of portrait information under the two card slots can be stretched by 2 points. In this way, the accuracy of portrait information clustering can be improved.

[0105] In S404, the server determines the first similarity as the target similarity.

[0106] As can be seen from the above scheme, when the correlation indicated by the correlation degree of the first card slot and the second card slot is greater than the preset correlation threshold, it indicates that the first card slot and the second card slot have a certain correlation. In this case, the similarity of the two pieces of portrait information of the two card slots in the first time period can be stretched accordingly, so as to improve the recall rate.

[0107] For example, when the similarity of the two pieces of portrait information is 90 points, the two pieces of portrait information can be determined as the portrait information of the same person. After the similarity comparison of the first portrait information and the second portrait information, the first similarity is 88 points. If the similarity is not stretched, the first portrait information and the second portrait information will be classified into two archives respectively, but if the first card slot and the second card slot are strongly correlated, the first similarity of 88 points can be stretched to the second similarity of 91 points after similarity stretching. Since 91 points are greater than 90 points, the two pieces of portrait information can be determined as the portrait information of the same person, and thus classified into the same archive. As can be seen from the example, the similarity of the two graphics is stretched based on the correlation of the card slot in the embodiments of the present application, which can reduce the phenomenon that multiple pieces of portrait information of a person are classified into multiple archives, so as to improve the recall rate.

[0108] The following Bayesian theory further describes the beneficial effects brought by the scheme provided in the embodiments of the present application:

[0109]

[0110] In formula (1):

[0111] X represents a threshold value;

[0112] A represents first portrait information collected by the first portal in a first time period;

[0113] B represents second portrait information collected by the second portal in the first time period;

[0114] P(A and B is a person) represents a probability of occurrence of an event “A and B is a person”; wherein the event “A and B is a person” represents an event that the first portrait information and the second portrait information are portrait information of the same person;

[0115] P(A and B is a person | X = μ1) represents a probability of occurrence of the event “A and B is a person” under the condition that the event “X = μ1” occurs;

[0116] P(X = μ1 | A and B is a person) represents a probability of occurrence of the event “X = μ1” under the condition that the event “A and B is a person” occurs;

[0117] P(X = μ1) represents a probability of occurrence of the event “X = μ1”.

[0118] When the degree of correlation between the first portal and the second portal indicates that the correlation is greater than a preset correlation threshold value, it indicates that the first portal and the second portal have a certain correlation, for example, the first portal and the second portal are strongly correlated, in this case, P(A and B is a person) will become larger, while P(X = μ1 | A and B is a person) and P(X = μ1) are constants and will not change. Further, P(A and B is a person | X = μ1) will also become larger, so that the threshold value on the fuzzy boundary can bring a larger probability of correct samples. Further, for the first portal and the second portal, when the two portals are strongly correlated, the probability that the two portrait information under the two portals is that of the same person will become larger.

[0119] It can be seen from the above Bayesian theory that the scheme provided in the embodiments of the present application is theoretically feasible, and the advantages brought by the embodiments of the present application have theoretical basis in the parameter setting of the link and are relatively effective. On the other hand, the case of multiple people in an archive can be described by the accuracy index, and the scheme provided in the embodiments of the present application can be used to compare the similarity of two portrait information, and if the similarity indicated by the similarity comparison result is greater than the similarity threshold, the two portrait information can be classified into the same archive. The scheme provided in the embodiments of the present application has no effect on this rule, that is, the scheme provided in the embodiments of the present application can not reduce the accuracy, so as to achieve the purpose of improving the recall rate without reducing the accuracy.

[0120] In another aspect, in S404, when the degree of correlation of the first and second card holes indicates that the correlation is not greater than the preset correlation threshold, it indicates that the first and second card holes are weakly correlated, and in this case, the probability that the two portrait information of the two card holes belongs to the same person is low. In this case, clustering can be performed only based on the similarity comparison result of the first and second portrait information, and the similarity of the portrait information is no longer stretched, so that the clustering of the portrait information can be more accurate.

[0121] It is worth noting that the above content is an example of two portrait information under two card holes, and in actual application, the similarity between any two pictures under any two card holes can be determined to obtain the corresponding target similarity between any two pictures, so as to obtain a similarity matrix, and then the pictures in all card holes can be clustered by using the similarity matrix. The determination scheme of the corresponding target similarity between any two pictures can be referred to the foregoing Figure 4 The provided scheme will not be described here.

[0122] Based on the above content, Figure 5 An example of another portrait information clustering method provided in the embodiments of the present application is shown. Figure 5 The scheme shown can be applied to a period of time after cold start. For example, it can be applied before step 201 of the foregoing Figure 2 , so as to accumulate a certain amount of portrait information in advance. The method can be performed by the server 101 in Figure 1 , or can be performed by a module, a chip or a unit in the server 101.

[0123] Please refer to Figure 5 , the method comprises:

[0124] S501, the server receives multiple portrait information collected by multiple card holes in a third time period.

[0125] In S501, the plurality of sockets can include a first socket and a second socket, and can also include other sockets; the plurality of portrait information can include at least two pieces of portrait information, and the following will be introduced by taking a third portrait information and a fourth portrait information included in the plurality of portrait information as an example. The third time period can be a historical time period before the first time period.

[0126] In S502, the server obtains feature information of each piece of portrait information by using a face analysis model.

[0127] In S503, the server can calculate a cosine similarity between the feature information corresponding to the two pieces of portrait information, to obtain a similarity between the two pieces of portrait information.

[0128] In S504, the server clusters the portrait information according to a preset similarity threshold specified by a computer, to obtain one or more portrait information clusters.

[0129] In S504, specifically, a plurality of modes such as hierarchical clustering can be used. For example, initialization can be performed, and then each sample is regarded as a cluster. Then, the distance between any two clusters is calculated, and the distance between two clusters can be calculated by using the nearest distance, the farthest distance or the weighted average distance of the points in the clusters. The two clusters with the nearest distance are found out, and the two clusters are merged. This step is repeated until the distance between the two farthest clusters exceeds the preset similarity threshold. In this way, a portrait information cluster is obtained, and the portrait information cluster in an ideal state refers to a cluster formed by the portrait information of a person captured by a plurality of sockets in a period of time.

[0130] In S505, the server classifies the obtained portrait information cluster into a base library.

[0131] In S505, for a portrait information cluster, the server can compare the portrait information cluster with an existing file in the base library. If the comparison shows that there is a file in the base library, and the similarity between the feature information corresponding to the portrait information included in the file and the feature information corresponding to the portrait information in the portrait information cluster is greater than a preset similarity threshold, the portrait information cluster can be classified into the file. Otherwise, a new file can be created in the base library.

[0132] In a possible implementation, in the case of cold start, that is, in the case that the base library does not have sufficient files, a certain number of files can be accumulated in the base library by the above-mentioned Figure 5 method. Then, the correlation between the sockets can be calculated by the foregoing Figure 3 scheme, and then the similarity between two pieces of portrait information is stretched based on the correlation between the sockets by the scheme shown in the above-mentioned Figure 2 or Figure 4 , so as to improve the recall rate of clustering.Figure 2 、 Figure 3 、 Figure 4 and Figure 5 The various schemes shown in

[0133] As shown in Figure 6 based on the same inventive concept, a personage clustering device 600 is provided, comprising an acquisition module 601 and a clustering module 602.

[0134] The acquisition module 601 is configured to acquire first personage information collected by a first lens in a first time period, and acquire second personage information collected by a second lens in the first time period.

[0135] The clustering module 602 is configured to determine a target similarity of the first personage information and the second personage information according to a correlation degree of the first lens and the second lens and a similarity comparison result of the first personage information and the second personage information. The correlation degree can be determined based on information of a person passing through the first lens and a person passing through the second lens in a second time period.

[0136] In some possible embodiments, the clustering module 602 is specifically configured to: perform similarity comparison on the first personage information and the second personage information to obtain a first similarity; and in a case where the correlation degree of the first lens and the second lens indicates that the correlation is greater than a preset correlation threshold, determine a second similarity corresponding to the first similarity as the target similarity according to a preset similarity mapping relationship, wherein the similarity mapping relationship comprises a corresponding relationship between the first similarity and the second similarity.

[0137] In some possible embodiments, the clustering module 602 is further configured to: in a case where the correlation degree of the first lens and the second lens indicates that the correlation is not greater than the preset correlation threshold, determine the first similarity as the target similarity.

[0138] Other related content of this scheme can be referred to the related content shown in the foregoing Figure 2 、 Figure 3 、 Figure 4 and Figure 5 , which will not be described here again.

[0139] After introducing the personage clustering method and device of the example embodiment of the present application, next, the electronic device according to another example embodiment of the present application is introduced.

[0140] Those skilled in the art can understand that each aspect of the present application can be implemented as a system, a method or a program product. Therefore, each aspect of the present application can be specifically implemented as follows: a complete hardware embodiment, a complete software embodiment (including firmware, microcode, etc.), or an embodiment combining hardware and software aspects, which can be collectively referred to as "circuitry", "module" or "system" here.

[0141] In some possible implementation, the electronic device according to the present application can at least include at least one processor, and at least one memory. Wherein, the memory stores program code, when the program code is executed by the processor, the processor executes the steps in the portrait clustering method according to various exemplary embodiments of the present application described above in the specification.

[0142] The electronic device 130 according to this embodiment of the present application will be described below with reference to Figure 7 Figure 7 The display electronic device 130 is only an example, and should not bring any limitation to the function and use range of the embodiments of the present application.

[0143] As Figure 7 shown, the electronic device 130 is in the form of a general electronic device. The components of the electronic device 130 can include but are not limited to: the above-mentioned at least one processor 131, the above-mentioned at least one memory 132, the bus 133 connecting different system components including the memory 132 and the processor 131.

[0144] The electronic device can be used to perform the following by an input / output (I / O) interface: acquiring first portrait information collected by a first portal in a first time period; acquiring second portrait information collected by a second portal in the first time period. According to the degree of correlation between the first portal and the second portal, and the similarity comparison result of the first portrait information and the second portrait information, determine the target similarity of the first portrait information and the second portrait information; wherein, the degree of correlation between the first portal and the second portal indicates the correlation, which is used to indicate the proportion of people passing through the second portal among people passing through the first portal in a second time period, and / or the proportion of people passing through the first portal among people passing through the second portal in the second time period; according to the target similarity, clustering the first portrait information and the second portrait information.

[0145] The bus 133 represents one or more of several types of bus structures, including a memory bus or a memory controller, a peripheral bus, a processor or a local bus using any of the bus structures.

[0146] ​The memory 132 can include a readable medium in the form of volatile memory, such as random access memory (RAM) 1321 and / or cache memory 1322, and can further include read only memory (ROM) 1323.

[0147] The memory 132 can also include a program / utility 1325 having a set of programs / modules 1324, including an operating system, one or more application programs, other program modules, and program data, each or some combination thereof, which may

[0148] The electronic device 130 can also communicate with one or more external devices 134 such as a keyboard or pointing device, through an I / O interface 135. Further, the electronic device 130 can communicate with one or more devices that enable user interaction with the electronic device 130, and / or one or more devices that enable communication of the electronic device 130 with one or more other electronic devices. This communication can be via the I / O interface 135. The electronic device 130 can also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or the Internet) through a network adapter 136. As depicted, the network adapter 136 communicates with the other components of the electronic device 130 via the bus 133. It should be appreciated that the network adapter 136 can also be connected to the other components of the electronic device 130 in some other fashion such as over a wireless connection, etc. Figure 7 Other hardware and / or software modules that can be used in conjunction with the electronic device 130 can also include, but are not limited to, microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data archival storage systems, etc., as non-limiting examples.

[0149] In some possible embodiments, various aspects of a person image clustering method provided by the present application can also be implemented as a program product, including a program code for causing a computer device to perform the steps of a person image clustering method according to various exemplary embodiments of the present application described above in the specification when the program product is run on the computer device.

[0150] A program product of embodiments of the present application for portrait clustering can employ a portable compact disc read-only memory (CD-ROM) and include a program code, and can be executed on an electronic device. However, the program product of the present application is not limited thereto, and in the present document, a readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0151] A program product of embodiments of the present application for portrait clustering can employ a portable compact disc read-only memory (CD-ROM) and include a program code, and can be executed on an electronic device. However, the program product of the present application is not limited thereto, and in the present document, a readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0152] A readable signal medium can include a data signal that is propagated in baseband or that is propagated as a carrier wave, in which a readable program code is embodied. Such a propagated signal can take any of a variety of forms, including but not limited to electro-magnetic, optical, or any suitable combination thereof. A readable signal medium can also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

[0153] Program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object-oriented programming language such as Java, C++, and the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on a user's electronic device, partly on the user's electronic device, as a stand-alone software package, partly on the user's electronic device and partly on a remote electronic device or entirely on the remote electronic device or server. In the latter scenario, the remote electronic device can be connected to the user's electronic device through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external electronic device (for example, through the Internet using an Internet Service Provider).

[0154] It should be noted that, although several units or sub-units of the apparatus are mentioned in the above detailed description, such a division is merely exemplary and not mandatory. Indeed, according to an embodiment of the application, the features and functionalities of two or more units described above can be embodied in one unit. Conversely, the features and functionalities of one unit described above can be further divided into units embodied by several units.

[0155] Moreover, although the operations of the method(s) herein can be described in a particular, sequential order, this order is not meant to be a limitation and is not intended to imply that

[0156] Those of skill in the art would understand that embodiments of the present application can be provided as a method, a system, or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, and the like) embodying computer readable program code.

[0157] The present application is described in reference to flow diagrams and / or block diagrams of methods, apparatus (systems) and computer program products according to this application. It will be understood that each block of the flow diagrams and / or block diagrams, and combinations of blocks in the flow diagrams and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flow diagrams and / or block diagrams block or blocks. Figure 1 The flow diagrams and / or block diagrams in the flow diagrams and / or block diagrams illustrate the functions and operations of processes which can be implemented as computer program instructions. Figure 1 The flow diagrams and / or block diagrams in the flow diagrams and / or block diagrams illustrate the functions and operations of processes which can be implemented as computer program instructions.

[0158] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flow diagrams and / or block diagrams. Figure 1 The flow diagrams and / or block diagrams in the flow diagrams and / or block diagrams illustrate the functions and operations of processes which can be implemented as computer program instructions. Figure 1 The flow diagrams and / or block diagrams in the flow diagrams and / or block diagrams illustrate the functions and operations of processes which can be implemented as computer program instructions.

[0159] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate computer-implemented processes, thus the instructions executed on the computer or other programmable data processing devices provide processes for implementing the functions specified in the flowchart Figure 1 one or more flows and / or blocks Figure 1 one or more blocks or steps of the functions specified in the flowchart

[0160] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A method of clustering portraits, characterized by, The method comprises: obtaining first portrait information collected by a first portal in a first time period; obtaining second portrait information collected by a second portal in the first time period; determining a target similarity between the first portrait information and the second portrait information according to a correlation degree of the first portal and the second portal and a similarity comparison result of the first portrait information and the second portrait information; the correlation degree is determined based on information of people passing through the first portal and people passing through the second portal in a second time period; the correlation degree comprises at least one of the following information: a first proportion of people passing through the second portal among people passing through the first portal in the second time period; a second proportion of people passing through the first portal among people passing through the second portal in the second time period; performing clustering on the first portrait information and the second portrait information according to the target similarity; wherein the determining of the target similarity comprises: performing similarity comparison on the first portrait information and the second portrait information to obtain a first similarity; in a case where the correlation indicated by the correlation degree of the first portal and the second portal is greater than a preset correlation threshold, determining a second similarity corresponding to the first similarity as the target similarity according to a similarity mapping relationship corresponding to the correlation degree of the first portal and the second portal; wherein the similarity mapping relationship comprises a corresponding relationship between the first similarity and the second similarity.

2. The method of claim 1, wherein, The correlation degree is determined according to at least one of the following: a first ratio of a first quantity to a second quantity; the first quantity is a quantity of people passing through the first portal and passing through the second portal in the second time period, and the second quantity is a quantity of people passing through the first portal in the second time period; a second ratio of the first quantity to a third quantity; the third quantity is a quantity of people passing through the second portal in the second time period; or information indicating an average of the first ratio and the second ratio.

3. The method of claim 2, wherein, The information indicating the average of the first ratio and the second ratio comprises at least one of the following: an average of the first ratio and the second ratio; or a harmonic average of the first ratio and the second ratio.

4. The method according to any one of claims 1 to 3, characterized in that, The similarity degree indicated by the second similarity is higher than the similarity degree indicated by the first similarity.

5. The method according to any one of claims 1 to 3, wherein After the similarity comparison on the first portrait information and the second portrait information to obtain the first similarity, the method further comprises: in a case where the correlation indicated by the correlation degree of the first portal and the second portal is not greater than a preset correlation threshold, determining the first similarity as the target similarity.

6. An apparatus for clustering portraits, the apparatus comprising: The apparatus comprises: an obtaining module, configured to obtain first portrait information collected by a first portal in a first time period; and obtain second portrait information collected by a second portal in the first time period; The clustering module is configured to: determine a target similarity between the first portrait information and the second portrait information according to a degree of correlation between the first passageway and the second passageway and a similarity comparison result of the first portrait information and the second portrait information; and cluster the first portrait information and the second portrait information according to the target similarity; the degree of correlation is determined based on information about a person passing through the first passageway and a person passing through the second passageway in a second time period; and the degree of correlation includes at least one of the following: a first proportion of the person passing through the second passageway among the person passing through the first passageway in the second time period; and a second proportion of the person passing through the first passageway among the person passing through the second passageway in the second time period. The clustering module is specifically configured to: perform a similarity comparison on the first portrait information and the second portrait information to obtain a first similarity; in a case where the degree of correlation between the first passageway and the second passageway is greater than a preset correlation threshold, determine a second similarity corresponding to the first similarity as the target similarity according to a similarity mapping relationship corresponding to the degree of correlation between the first passageway and the second passageway; and the similarity mapping relationship includes a corresponding relationship between the first similarity and the second similarity.

7. The apparatus of claim 6, wherein, The degree of correlation is determined according to at least one of the following: a first ratio of a first quantity to a second quantity; the first quantity is a quantity of a person passing through the first passageway and the second passageway in the second time period, and the second quantity is a quantity of a person passing through the first passageway in the second time period; a second ratio of the first quantity to a third quantity; the third quantity is a quantity of a person passing through the second passageway in the second time period; or information indicating a mean value of the first ratio and the second ratio.

8. The apparatus of claim 7, wherein, The information indicating the mean value of the first ratio and the second ratio includes at least one of the following: an average of the first ratio and the second ratio; or a harmonic mean of the first ratio and the second ratio.

9. The device of any one of claims 6-8, wherein, The second similarity indicates a higher degree of similarity than the first similarity.

10. The apparatus of any one of claims 6-8, wherein, The clustering module is further configured to: in a case where the degree of correlation between the first passageway and the second passageway is not greater than a preset correlation threshold, determine the first similarity as the target similarity.

11. An electronic device, comprising: The electronic device includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the at least one processor, by executing the instructions stored in the memory, causes the electronic device to perform the method of any one of claims 1-5.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium is configured to store instructions that, when executed, cause the method of any one of claims 1-5 to be implemented.

Images