An archival filing method and related apparatus

Abstract

Embodiments of the present application provide an archive filing method and related equipment, the method comprising: screening an abnormal human body archive set from a to-be-processed archive, the abnormal human body archive set comprising a plurality of abnormal human body archives; based on a face archive image and a human body archive image in the abnormal human body archive set, performing splitting processing on the plurality of abnormal human body archives to obtain split human body archives; based on the split human body archives, performing re-archiving on other human body archives in the to-be-processed archive that do not belong to the split human body archives to obtain an updated archiving result. In this way, the effectiveness of the archive is inferred based on human body information rather than face information, which can avoid the influence of similar faces, side faces, wearing masks and the like, the human body can provide more rich auxiliary information, thereby increasing the number of effective archives and improving the archive filing accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of archive filing, and in particular to an archive filing method and related equipment. BACKGROUND

[0002] The existing picture filing method usually extracts the facial feature information of the person in the snapshot picture, calculates the similarity between two snapshot pictures according to the facial feature information, and determines whether to classify the two snapshot pictures into the same archive according to the size of the similarity. When the shooting angle of the face changes or the facial features are blurred, the reliability of the calculated similarity is low, resulting in poor picture filing effect. The body image captured also needs to be filed, and the filing accuracy of the body in the absence of a face is also very low. The current filing algorithm performs clustering filing based on social group detection, and the existence of error edges caused by low-quality pictures or occlusion leads to the situation that multiple people are classified into one archive in the final result set, thereby reducing the effectiveness and accuracy of the archive. SUMMARY

[0003] Embodiments of the present application provide an archive filing method, aiming to solve the problem of the existing clustering result that multiple people are classified into one archive, and reduce the effectiveness and accuracy of the archive.

[0004] In a first aspect, embodiments of the present application provide an archive filing method, which comprises:

[0005] Screening an abnormal body archive set from the to-be-processed archive, wherein the abnormal body archive set comprises a plurality of abnormal body archives;

[0006] Splitting processing a plurality of abnormal body archives based on the face archive images and body archive images in the abnormal body archive set, to obtain split body archives;

[0007] Based on the split body archives, re-filing other body archives in the to-be-processed archive that do not belong to the split body archives, to obtain an updated filing result.

[0008] Optionally, the step of screening an abnormal body archive set from the to-be-processed archive comprises:

[0009] Screening a body archive with a body archive image quantity greater than a first preset threshold from the to-be-processed archive as a to-be-processed body archive;

[0010] If the number of to-be-processed body archives is multiple, determining the multiple to-be-processed body archives as multiple abnormal body archives to constitute an abnormal body archive set of the to-be-processed archive.

[0011] Optionally, the splitting processing of the plurality of abnormal human body archives based on the face archive images and the human body archive images in the plurality of abnormal human body archive sets, to obtain the split human body archives, comprises:

[0012] For each of the abnormal human body archives, the first face feature similarity corresponding to any two different face archive images in the abnormal human body archive is calculated, and the first human body feature similarity corresponding to any two different human body archive images in the abnormal human body archive is calculated;

[0013] If the average similarity of the first face feature similarity is lower than the first face feature similarity threshold and the average similarity of the first human body feature similarity is lower than the preset first human body feature similarity threshold, the splitting processing of the plurality of abnormal human body archives based on the second face feature similarity between the face archive images contained in the plurality of abnormal human body archives and the second human body feature similarity between the human body archive images contained in the plurality of abnormal human body archives, to obtain the split human body archives.

[0014] Optionally, the splitting processing of the plurality of abnormal human body archives based on the second face feature similarity between the face archive images contained in the plurality of abnormal human body archives and the second human body feature similarity between the human body archive images contained in the plurality of abnormal human body archives, to obtain the split human body archives, comprises:

[0015] If the second face feature similarity is lower than the second face feature similarity threshold and the second human body feature similarity is lower than the second human body feature similarity threshold, the corresponding abnormal human body archives in the plurality of abnormal human body archives are split to obtain the split human body archives.

[0016] Optionally, the re-archiving of the other human body archives in the to-be-processed archives which do not belong to the split human body archives based on the split human body archives, to obtain the updated archiving result, comprises:

[0017] If there is a face archive image in the other human body archives in the to-be-processed archives which do not belong to the split human body archives, the matching is preferentially performed according to the face archive image, and the other human body archives are classified into the human body archive in which the number of times of satisfying the preset face similarity threshold is the most among the split human body archives;

[0018] If there is no face archive image in the other human body archives in the to-be-processed archives which do not belong to the split human body archives, the matching is performed according to the human body archive image, and the other human body archives are classified into the human body archive in which the number of times of satisfying the preset human body similarity threshold is the most among the split human body archives.

[0019] Optionally, before the step of screening the abnormal human body archive set from the to-be-processed archives, the method further comprises:

[0020] obtaining to-be-archived images, the to-be-archived images comprising to-be-archived face images and to-be-archived human body images;

[0021] performing clustering processing on the to-be-archived face images in the to-be-archived images to obtain face clustering results, and performing clustering processing on the to-be-archived human body images in the to-be-archived images to obtain human body clustering results, the face clustering results comprising at least one face archive, and the human body clustering results comprising at least one human body archive;

[0022] associating the face clustering results and the human body clustering results to obtain the to-be-processed archives.

[0023] In a second aspect, an archive archiving apparatus is also provided, which comprises:

[0024] a screening module configured to screen an abnormal human body archive set from to-be-processed archives, the abnormal human body archive set comprising a plurality of abnormal human body archives;

[0025] a splitting module configured to split a plurality of the abnormal human body archives based on face archive images and human body archive images in the abnormal human body archive set to obtain split human body archives;

[0026] a re-archiving module configured to re-archive other human body archives in the to-be-processed archives that do not belong to the split human body archives based on the split human body archives to obtain an updated archiving result.

[0027] Optionally, the screening module comprises:

[0028] a screening unit configured to screen human body archives with a number of human body archive images greater than a first preset threshold from the to-be-processed archives as to-be-processed human body archives;

[0029] a determination unit configured to determine, if a number of the to-be-processed human body archives is a plurality, that the plurality of to-be-processed human body archives are a plurality of abnormal human body archives to constitute the abnormal human body archive set of the to-be-processed archives.

[0030] In a third aspect, an electronic device is provided, which comprises a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and the processor implements the steps in the archive archiving method provided by the embodiments of the present application when executing the computer program.

[0031] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, and the computer readable storage medium stores a computer program. The computer program is executed by a processor to implement the steps of the file archiving method provided by the embodiment of the present application.

[0032] In the embodiment of the present application, the abnormal human body file set is screened out from the to-be-processed files, the abnormal human body file set includes a plurality of abnormal human body files, the plurality of abnormal human body files are split based on the face file images and the human body file images in the abnormal human body file set to obtain split human body files, and other human body files in the to-be-processed files that do not belong to the split human body files are re-archived based on the split human body files to obtain an updated archiving result. In this way, the effectiveness of the file is inferred based on the human body information instead of the face information, the influence of similar faces or side faces, wearing masks and the like can be avoided, the human body can provide more rich auxiliary information, and then the number of effective files is increased and the archiving precision of the file is improved. BRIEF DESCRIPTION OF DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0034] Figure 1 is a flowchart of a file archiving method provided by an embodiment of the present application;

[0035] Figure 2 is Figure 1 is a method flowchart provided in step 101 in the embodiment;

[0036] Figure 3 is a structural schematic diagram of a file archiving device provided by an embodiment of the present application;

[0037] Figure 4 is Figure 3 is a structural schematic diagram of a screening module provided in the embodiment;

[0038] Figure 5 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0039] With reference to the drawings and embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0040] As shown in Figure 1 , Figure 1 is a file archiving method flowchart provided by an embodiment of the present application. The file archiving method comprises the following steps:

[0041] Step 101, screening an abnormal human body file set from the to-be-processed files.

[0042] The to-be-processed files are files obtained by archiving to-be-archived images according to a preset clustering algorithm in advance, and the files need to be secondarily archived. The to-be-processed files can also refer to files that have been archived once. The to-be-processed files can be one or multiple.

[0043] The abnormal human body file set comprises multiple abnormal human body files.

[0044] The abnormal human body file set is a set of all abnormal human body files screened from the to-be-processed files. Of course, the abnormal human body file set comprises at least two abnormal human body files. Of course, when there are multiple abnormal human body files in the to-be-processed files, the abnormal human body file set also comprises multiple abnormal human body files.

[0045] The to-be-processed files comprise a face file and multiple human body files corresponding to the face file. Each human body file comprises multiple human body file images. The face file comprises multiple face file images.

[0046] Specifically, as shown in Figure 2 , step 101 comprises:

[0047] Step 201, screening a human body file with a file image quantity greater than a first preset threshold from the to-be-processed files as a to-be-processed human body file.

[0048] The first preset threshold is a condition for judging whether a file is a to-be-processed human body file, which is set in advance.

[0049] Specifically, the number of body archive images of each body archive corresponding to the face archive in the to-be-processed archive is compared with the first preset threshold, to determine whether the number of body archive images of each body archive is greater than the first preset threshold. If the number of body archive images of the body archive in the to-be-processed archive is greater than the first preset threshold, it is indicated that the body archive of the to-be-processed archive meets the requirement of the to-be-processed body archive. After all the body archives in the to-be-processed archive are traversed, all the body archives in the to-be-processed archive whose number of body archive images meets the first preset threshold are filtered out, and then the corresponding to-be-processed body archive is obtained. For example, the to-be-processed archive A includes one face archive and three body archives, wherein the number of body archive images of the three body archives is 100, 90 and 3 respectively. When the first preset threshold is set to be greater than 80, the number of body archive images of the body archive with the number of 3 in the three body archives of the to-be-processed archive A is less than the first preset threshold, so the body archive with the number of 3 is filtered out, and only the body archives with the number of 100 and 90 are left. Therefore, the to-be-processed body archive in the to-be-processed archive A is the body archives with the number of 100 and 90. At this time, the number of archives of the to-be-processed body archive is two.

[0050] Step 202, if the number of to-be-processed body archives is multiple, the multiple to-be-processed body archives are determined as multiple abnormal body archives to constitute an abnormal body archive set of the to-be-processed archive.

[0051] Specifically, the number of to-be-processed body archives is compared with 1, and if the number of to-be-processed body archives is greater than 1, it is indicated that the number of to-be-processed body archives is multiple.

[0052] It should be noted that, in theory, a high-quality archive includes one face archive and one body archive corresponding to the face archive. When one face archive in an archive corresponds to multiple body archives, it is indicated that the archive is an abnormal archive, and the archive needs to be secondarily archived.

[0053] More specifically, when multiple to-be-processed body archives are filtered out from the to-be-processed archive, it is indicated that the to-be-processed archive has multiple abnormal body archives, and the multiple abnormal body archives constitute an abnormal body archive set of the to-be-processed archive.

[0054] Step 102, based on the face archive image and the body archive image in the abnormal body archive set, the multiple abnormal body archives are split to obtain split body archives.

[0055] Specifically, step 102 includes:

[0056] For each abnormal human archive, the first face feature similarity corresponding to any two different face archive images in the abnormal human archive is calculated, and the first human feature similarity corresponding to any two different human archive images in the abnormal human archive is calculated; if the average similarity of the first face feature similarity is lower than the first face feature similarity threshold and the average similarity of the first human feature similarity is lower than the preset first human feature similarity threshold, then the multiple abnormal human archives are split based on the second face feature similarity between the face archive images contained in the multiple abnormal human archives and the second human feature similarity between the human archive images contained in the multiple abnormal human archives, to obtain the split human archives.

[0057] More specifically, after determining the abnormal human archive, the first face feature similarity corresponding to any two different face archive images in the abnormal human archive is calculated, and the first human feature similarity corresponding to any two different human archive images in the abnormal human archive is calculated, and then the average similarity of all first face feature similarities and the average similarity of all first human feature similarities are calculated, and then compared with the first face feature similarity threshold and the first human feature similarity threshold respectively. If the average similarity of the first face feature similarity is lower than the first face feature similarity threshold and the average similarity of the first human feature similarity is lower than the preset first human feature similarity threshold, then the second face feature similarity between the face archive images contained in the multiple abnormal human archives and the second human feature similarity between the human archive images contained in the multiple abnormal human archives are calculated, and then the multiple abnormal human archives are split to obtain the split human archives.

[0058] Further, based on the second face feature similarity between the face archive images contained in the multiple abnormal human archives and the second human feature similarity between the human archive images contained in the multiple abnormal human archives, the multiple abnormal human archives are split to obtain the split human archives, including:

[0059] If the second face feature similarity is lower than the second face feature similarity threshold and the second human feature similarity is lower than the second human feature similarity threshold, then the corresponding abnormal human archives in the multiple abnormal human archives are split to obtain the split human archives. In this way, a secondary matching can be performed on all abnormal human archives, further increasing the number of effective archives and improving the archive filing accuracy.

[0060] Step 103, based on the split human archives, the other human archives in the to-be-processed archives that do not belong to the split human archives are re-archived to obtain an updated archiving result.

[0061] Specifically, step 103 includes:

[0062] If there is a face archive image in other human body archives in the to-be-processed archives which do not belong to the split human body archives, the face archive image is preferentially matched, and is classified into the human body archive in which the number of times that the face similarity meets the preset face similarity threshold is the most. If there is no face archive image in other human body archives in the to-be-processed archives which do not belong to the split human body archives, the human body archive image is matched, and is classified into the human body archive in which the number of times that the human body similarity meets the preset human body similarity threshold is the most.

[0063] The other human body archives in the to-be-processed archives which do not belong to the split human body archives refer to the human body archives in the to-be-processed archives in which the number of human body archive images is less than a first preset threshold.

[0064] Specifically, first, the face similarity of the face archive image in the other human body archives in the to-be-processed archives which do not belong to the split human body archives is calculated with the face archive image in each human body archive in the split human body archives, and is compared with the preset face similarity threshold. The face archive image in the other human body archives in the to-be-processed archives which do not belong to the split human body archives is merged with the human body archive in which the number of times that the face similarity of the face archive image in the split human body archives meets the preset face similarity threshold is the most, to obtain a new archive archive.

[0065] Of course, if there is no face archive image in the other human body archives in the to-be-processed archives which do not belong to the split human body archives, the human body similarity of the human body archive image in the other human body archives in the to-be-processed archives which do not belong to the split human body archives is calculated with the human body archive image in each human body archive in the split human body archives, and is compared with the preset human body similarity threshold. The human body archive image in the other human body archives in the to-be-processed archives which do not belong to the split human body archives is merged with the human body archive in which the number of times that the human body similarity of the human body archive image in the split human body archives meets the preset human body similarity threshold is the most, to obtain a new archive archive.

[0066] In the embodiment of the application, the archive archiving method further comprises:

[0067] Obtaining a to-be-archived image, the to-be-archived image including a to-be-archived face image and a to-be-archived human body image.

[0068] The to-be-archived image can be an image that needs to be archived by a user. The to-be-archived image can be an image including a face image, a human body image, or both a face image and a human body image. The number of to-be-archived images is multiple.

[0069] Specifically, the image to be archived can be obtained from a specific image database, can be obtained by tracking and capturing personnel in the field of view of a front-end device, or can be obtained from a remote connection. The front-end device can be a camera, a camera, or the like. Of course, the image to be archived can be the optimal face and body information selected and reported during tracking. The received track by the cloud can include a face, a body, or both a face and a body.

[0070] The image to be archived in the image to be archived is clustered to obtain a face clustering result, and the image to be archived in the image to be archived is clustered to obtain a body clustering result. The face clustering result includes at least one face archive, and the body clustering result includes at least one body archive.

[0071] The image to be archived in the image to be archived is clustered to obtain a face clustering result, and the image to be archived in the image to be archived is clustered to obtain a body clustering result. The face clustering result includes at least one face archive, and the body clustering result includes at least one body archive.

[0072] The image to be archived in the image to be archived is clustered to obtain a face clustering result, and the image to be archived in the image to be archived is clustered to obtain a body clustering result. The face clustering result includes at least one face archive, and the body clustering result includes at least one body archive.

[0073] The face archive is an archive obtained by archiving the face image to be archived. In theory, a face archive can only include face images of the same person. The body archive is an archive obtained by archiving the body image to be archived. Of course, in theory, a body archive includes multiple body images of the same person.

[0074] Specifically, after obtaining the to-be-archived image, when the to-be-archived image includes a to-be-archived face image, face feature search is performed on all faces in the to-be-archived image, and face similarity between each face is calculated respectively. The face ID that satisfies the face similarity threshold requirement is taken as a node, a graph is built with the face similarity as an edge, face clustering archiving is performed through the infoMap algorithm, and finally N face clustering result sets are obtained. The face clustering result sets are iterated, each face clustering result set is a new face archive, a new PID is added as an archive ID, and the archive ID of the face result set corresponding to the face track (face image) is updated to the PID.

[0075] Of course, when the to-be-archived image includes a to-be-archived body image, body feature search is performed on all received to-be-archived body images, body similarity between each body is calculated respectively, similar bodies are obtained, the body ID that satisfies the body similarity threshold requirement is taken as a node, a graph is built with the body similarity as an edge, body clustering archiving is performed through the infoMap algorithm, and finally N body clustering result sets are obtained. The body clustering result sets are iterated, each body clustering result set is a new body set ID, denoted as BID, and the body set ID of the body track corresponding to the body clustering result set is updated to the BID.

[0076] The face clustering result and the body clustering result are associated to obtain a to-be-processed archive.

[0077] Specifically, after archiving the to-be-archived face image and the to-be-archived body image respectively, the face archive corresponding to the image of the body clustering result is obtained, the archive ID of the face archive with the most records is taken as the archive ID of the associated archive, the archive ID of the body track corresponding to the body clustering result is updated to the PID, and the association between the face clustering result and the body clustering result is realized to obtain a final archive set, and then a to-be-processed archive is obtained. The most records refer to the most face similarity times after iterating all face archives.

[0078] In the embodiment of the application, an abnormal body archive set is screened out from the to-be-processed archive, the abnormal body archive set includes a plurality of abnormal body archives; the plurality of abnormal body archives are processed through splitting based on face archive images and body archive images in the abnormal body archive set, to obtain split body archives; other body archives in the to-be-processed archive that do not belong to the split body archives are re-archived based on the split body archives, to obtain an updated archiving result. This method uses body information instead of face information to speculate the effectiveness of the archive, which can avoid the influence of face similarity or side face, wearing a mask, etc., the body can provide more rich auxiliary information, thereby increasing the number of effective archives and improving the archiving accuracy.

[0079] AsFigure 3 As shown in the figure, the archive filing device 300 comprises:

[0080] a screening module 301, configured to screen a set of abnormal human body archives from the archives to be processed, the set of abnormal human body archives comprising a plurality of abnormal human body archives;

[0081] a splitting module 302, configured to split the plurality of abnormal human body archives based on the face archive images and the human body archive images in the set of abnormal human body archives, to obtain split human body archives;

[0082] a re-filing module 303, configured to re-file other human body archives in the archives to be processed that do not belong to the split human body archives based on the split human body archives, to obtain an updated filing result.

[0083] Optionally, as shown in the figure, Figure 4 the screening module 301 comprises:

[0084] a screening unit 3011, configured to screen human body archives with a number of face archive images greater than a first preset threshold from the archives to be processed as human body archives to be processed;

[0085] a determination unit 3012, configured to, if the number of the human body archives to be processed is a plurality, determine the plurality of human body archives to be processed as a plurality of abnormal human body archives to constitute the set of abnormal human body archives of the archives to be processed.

[0086] Optionally, the splitting module 302 comprises:

[0087] a calculation unit, configured to, for each abnormal human body archive, calculate a first face feature similarity corresponding to any two different face archive images in the abnormal human body archive, and calculate a first human body feature similarity corresponding to any two different human body archive images in the abnormal human body archive;

[0088] a splitting unit, configured to, if an average similarity of the first face feature similarity is lower than a first face feature similarity threshold and an average similarity of the first human body feature similarity is lower than a preset first human body feature similarity threshold, split the plurality of abnormal human body archives based on a second face feature similarity between the face archive images contained in the plurality of abnormal human body archives and a second human body feature similarity between the human body archive images contained in the plurality of abnormal human body archives, to obtain the split human body archives.

[0089] Optionally, the splitting unit is further configured to, if the second face feature similarity is lower than a second face feature similarity threshold and the second human body feature similarity is lower than a second human body feature similarity threshold, split the corresponding abnormal human body archives in the plurality of abnormal human body archives, to obtain the split human body archives.

[0090] Optionally, the re-archiving module 303 comprises:

[0091] The first matching unit is configured to, if there is a face archive image in other human body archives in the to-be-processed archives which do not belong to the split human body archives, preferentially perform matching according to the face archive image and is classified into the human body archive in which the number of times that the face similarity meets the preset face similarity threshold is the largest among the split human body archives;

[0092] The second matching unit is configured to, if there is no face archive image in other human body archives in the to-be-processed archives which do not belong to the split human body archives, perform matching according to the human body archive image and is classified into the human body archive in which the number of times that the human body similarity meets the preset human body similarity threshold is the largest among the split human body archives.

[0093] Optionally, before the screening module 301, the archive archiving device 300 further comprises:

[0094] The acquisition module is configured to acquire to-be-archived images, wherein the to-be-archived images comprise to-be-archived face images and to-be-archived human body images;

[0095] The clustering module is configured to perform clustering processing on the to-be-archived face images in the to-be-archived images to obtain a face clustering result, and perform clustering processing on the to-be-archived human body images in the to-be-archived images to obtain a human body clustering result, wherein the face clustering result comprises at least one face archive, and the human body clustering result comprises at least one human body archive;

[0096] The association module is configured to associate the face clustering result and the human body clustering result to obtain to-be-processed archives.

[0097] The archive archiving device 300 provided by the embodiment of the present application can realize each process of the archive archiving method in the method embodiment, and can achieve the same beneficial effects. To avoid repetition, details are not described herein.

[0098] Referring to Figure 5 , Figure 5 is a structural schematic diagram of an electronic device provided by the embodiment of the present application, as shown in Figure 5 , comprising a memory 402, a processor 401 and a computer program of an archive archiving method stored in the memory 402 and capable of running on the processor 401, wherein:

[0099] The processor 401 is configured to call the computer program stored in the memory 402 to perform the following steps:

[0100] screening an abnormal human body archive set from the to-be-processed archives, wherein the abnormal human body archive set comprises a plurality of abnormal human body archives;

[0101] The multiple abnormal human body archives are split based on the face archive images and the human body archive images in the abnormal human body archive set, to obtain split human body archives.

[0102] Other human body archives in the to-be-processed archives that do not belong to the split human body archives are re-archived based on the split human body archives, to obtain an updated archiving result.

[0103] Optionally, the processor 401 performs filtering of the abnormal human body archive set from the to-be-processed archives, including:

[0104] A human body archive with a number of human body archive images greater than a first preset threshold is filtered from the to-be-processed archives as a to-be-processed human body archive.

[0105] If the number of to-be-processed human body archives is multiple, the multiple to-be-processed human body archives are determined as multiple abnormal human body archives to constitute the abnormal human body archive set of the to-be-processed archives.

[0106] Optionally, the processor 401 performs split processing of the multiple abnormal human body archives based on the face archive images and the human body archive images in the multiple abnormal human body archive set, to obtain split human body archives, including:

[0107] For each abnormal human body archive, a first face feature similarity corresponding to any two different face archive images in the abnormal human body archive is calculated, and a first human body feature similarity corresponding to any two different human body archive images in the abnormal human body archive is calculated.

[0108] If the average similarity of the first face feature similarity is lower than a first face feature similarity threshold and the average similarity of the first human body feature similarity is lower than a preset first human body feature similarity threshold, split processing of the multiple abnormal human body archives is performed based on a second face feature similarity between the face archive images contained in the multiple abnormal human body archives and a second human body feature similarity between the human body archive images contained in the multiple abnormal human body archives, to obtain split human body archives.

[0109] Optionally, the processor 401 performs split processing of the multiple abnormal human body archives based on the second face feature similarity between the face archive images contained in the multiple abnormal human body archives and the second human body feature similarity between the human body archive images contained in the multiple abnormal human body archives, to obtain split human body archives, including:

[0110] If the second face feature similarity is lower than a second face feature similarity threshold and the second human body feature similarity is lower than a second human body feature similarity threshold, corresponding abnormal human body archives in the multiple abnormal human body archives are split, to obtain split human body archives.

[0111] Optionally, the processor 401 performs, based on the split human archives, re-archiving other human archives in the to-be-processed archive that do not belong to the split human archives, to obtain an updated archiving result, including:

[0112] If there is a face archive image in the other human archives in the to-be-processed archive that do not belong to the split human archives, the face archive image is preferentially matched and classified into the human archive in which the number of times that the face similarity meets a preset face similarity threshold is the largest.

[0113] If there is no face archive image in the other human archives in the to-be-processed archive that do not belong to the split human archives, the human archive image is matched and classified into the human archive in which the number of times that the human similarity meets a preset human similarity threshold is the largest.

[0114] Optionally, before the processor 401 filters the set of abnormal human archives from the to-be-processed archive, the processor 401 further performs:

[0115] Obtaining a to-be-archived image, the to-be-archived image including a to-be-archived face image and a to-be-archived human image;

[0116] Performing clustering processing on the to-be-archived face image in the to-be-archived image to obtain a face clustering result, and performing clustering processing on the to-be-archived human image in the to-be-archived image to obtain a human clustering result, the face clustering result including at least one face archive and the human clustering result including at least one human archive;

[0117] Associating the face clustering result and the human clustering result to obtain the to-be-processed archive.

[0118] It should be noted that the electronic device 400 provided by the embodiment of the present application can be applied to a smart phone, a computer, a server and the like that can perform the archive archiving method.

[0119] The electronic device 400 provided by the embodiment of the present application can implement each process of the archive archiving method in the above method embodiment, and can achieve the same beneficial effects. To avoid repetition, details are not repeated here.

[0120] The embodiment of the present application further provides a computer readable storage medium, and the computer readable storage medium stores a computer program. When the computer program is executed by a processor, each process of the archive archiving method or the application end archive archiving method provided by the embodiment of the present application is implemented, and the same technical effects can be achieved. To avoid repetition, details are not repeated here.

[0121] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing relevant hardware, and the program can be stored in a computer readable storage medium. When the program is executed, the program can include the processes of the above-mentioned embodiment methods. The storage medium can be a magnetic disc, an optical disc, a read-only memory (ROM) or a random access memory (RAM).

[0122] The above only describes the preferred embodiments of the present application, and cannot limit the scope of the present application. Any equivalent changes made according to the claims of the present application are still within the scope of the present application.

Claims

1. An archival filing method, characterized by, The method comprises the following steps: Screening an abnormal human archive set from the to-be-processed archive, wherein the abnormal human archive set comprises a plurality of abnormal human archives; Based on the face archive images and the human archive images in the abnormal human archive set, the plurality of abnormal human archives are split to obtain split human archives, comprising: for each abnormal human archive, calculating the first face feature similarity corresponding to any two different face archive images in the abnormal human archive, and calculating the first human feature similarity corresponding to any two different human archive images in the abnormal human archive; if the average similarity of the first face feature similarity is lower than a first face feature similarity threshold and the average similarity of the first human feature similarity is lower than a preset first human feature similarity threshold, then based on the second face feature similarity between the face archive images contained in the plurality of abnormal human archives and the second human feature similarity between the human archive images contained in the plurality of abnormal human archives, the plurality of abnormal human archives are split to obtain split human archives; The method comprises the following steps: Based on the face archive images and the human archive images in the abnormal human archive set, the plurality of abnormal human archives are split to obtain split human archives, comprising: if the second face feature similarity is lower than a second face feature similarity threshold and the second human feature similarity is lower than a second human feature similarity threshold, then the corresponding abnormal human archives in the plurality of abnormal human archives are split to obtain split human archives; 2. The method of claim 1, wherein, Based on the split human archives, other human archives in the to-be-processed archive that do not belong to the split human archives are re-archived to obtain an updated archiving result, comprising: if there are face archive images in the other human archives in the to-be-processed archive that do not belong to the split human archives, then the matching is preferentially performed according to the face archive images, and the other human archives are classified into the human archive with the most number of times of meeting the preset face similarity threshold in the split human archives; if there are no face archive images in the other human archives in the to-be-processed archive that do not belong to the split human archives, then the matching is performed according to the human archive images, and the other human archives are classified into the human archive with the most number of times of meeting the preset human similarity threshold in the split human archives. The method further comprises the following steps before the step of screening the abnormal human archive set from the to-be-processed archive: Screening a human archive with a number of human archive images greater than a first preset threshold from the to-be-processed archive as a to-be-processed human archive; 3. The method of claim 1, wherein, If the number of the to-be-processed human archives is a plurality, then the plurality of to-be-processed human archives are determined as a plurality of abnormal human archives to constitute the abnormal human archive set of the to-be-processed archive. Obtaining to-be-archived images, the to-be-archived images comprising to-be-archived face images and to-be-archived body images; Clustering the to-be-archived face images in the to-be-archived images to obtain face clustering results, and clustering the to-be-archived body images in the to-be-archived images to obtain body clustering results, the face clustering results comprising at least one face archive, and the body clustering results comprising at least one body archive; Associating the face clustering results and the body clustering results to obtain the to-be-processed archives.

4. An archival filing device, characterized by, The archive archiving apparatus comprises: A screening module configured to screen an abnormal body archive set from the to-be-processed archives, the abnormal body archive set comprising a plurality of abnormal body archives; A splitting module configured to split the plurality of abnormal body archives based on face archive images and body archive images in the abnormal body archive set to obtain split body archives, comprising: for each abnormal body archive, calculating a first face feature similarity corresponding to any two different face archive images in the abnormal body archive, and calculating a first body feature similarity corresponding to any two different body archive images in the abnormal body archive; if an average similarity of the first face feature similarity is lower than a first face feature similarity threshold and an average similarity of the first body feature similarity is lower than a preset first body feature similarity threshold, splitting the plurality of abnormal body archives based on a second face feature similarity between the face archive images contained in the plurality of abnormal body archives and a second body feature similarity between the body archive images contained in the plurality of abnormal body archives to obtain split body archives; The splitting the plurality of abnormal body archives based on the second face feature similarity between the face archive images contained in the plurality of abnormal body archives and the second body feature similarity between the body archive images contained in the plurality of abnormal body archives to obtain split body archives, comprising: if the second face feature similarity is lower than a second face feature similarity threshold and the second body feature similarity is lower than a second body feature similarity threshold, splitting corresponding abnormal body archives in the plurality of abnormal body archives to obtain split body archives; The splitting the plurality of abnormal body archives based on the second face feature similarity between the face archive images contained in the plurality of abnormal body archives and the second body feature similarity between the body archive images contained in the plurality of abnormal body archives to obtain split body archives, comprising: if the second face feature similarity is lower than a second face feature similarity threshold and the second body feature similarity is lower than a second body feature similarity threshold, splitting corresponding abnormal body archives in the plurality of abnormal body archives to obtain split body archives; The re-archiving module is configured to re-archive other human body archives in the to-be-processed archives that do not belong to the split human body archives based on the split human body archives, to obtain an updated archiving result, including: if there is a face archive image in the other human body archives in the to-be-processed archives that do not belong to the split human body archives, preferentially performing matching according to the face archive image, and being classified into a human body archive in which a face similarity meets a preset face similarity threshold value most frequently; and if there is no face archive image in the other human body archives in the to-be-processed archives that do not belong to the split human body archives, performing matching according to a human body archive image, and being classified into a human body archive in which a human body similarity meets a preset human body similarity threshold value most frequently.

5. The apparatus of claim 4, wherein, The screening module includes: The screening unit is configured to screen a human body archive with a number of human body archive images greater than a first preset threshold value from the to-be-processed archives as a to-be-processed human body archive. The determining unit is configured to, if the number of the to-be-processed human body archives is a plurality, determine the plurality of to-be-processed human body archives as a plurality of abnormal human body archives to constitute an abnormal human body archive set of the to-be-processed archives.

6. An electronic device, comprising: The computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the steps in the archiving method according to any one of claims 1 to 3. The computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the steps in the archiving method according to any one of claims 1 to 3.

7. A computer readable storage medium characterized in that, ​

Images