CN114519879A - Human body data archiving method, device, equipment and storage medium - Google Patents
Human body data archiving method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN114519879A CN114519879A CN202111681912.6A CN202111681912A CN114519879A CN 114519879 A CN114519879 A CN 114519879A CN 202111681912 A CN202111681912 A CN 202111681912A CN 114519879 A CN114519879 A CN 114519879A
- Authority
- CN
- China
- Prior art keywords
- human body
- body data
- face
- data
- human
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Processing Or Creating Images (AREA)
Abstract
The application relates to a human body data archiving method, a human body data archiving device, human body data archiving equipment and a storage medium, in particular to the technical field of computer vision. The method comprises the following steps: acquiring at least two target images and carrying out face clustering operation to obtain face aggregation files; classifying first human body data meeting a first similar condition with associated human body data of the face aggregation files in the target image into the face aggregation files aiming at each face aggregation file; and aiming at each updated face aggregation file, when the human body data meeting a second similar condition with the second human body data in the target image is searched in the updated face aggregation files, classifying the second human body data into the updated face aggregation files. The human body data are secondarily judged by the scheme, incomplete classification caused by poor data quality of the human body data related to the human face is avoided as much as possible, and the recall rate of filing the human body data into the human face aggregation file is improved.
Description
Technical Field
The invention relates to the field of computer vision, in particular to a human body data archiving method, a human body data archiving device, human body data archiving equipment and a storage medium.
Background
In the field of video monitoring, people information in a video image is generally required to be identified, and event information, such as in-out information and the like, generated by people in a target area is counted.
When the person information in the video image is identified, each person usually appears many times, and in order to accurately count the identity information and the occurrence event of each person appearing in the video image, a person skilled in the art usually adopts a face data archive gathering mode to classify similar faces into the same archive, and then classify similar persons into an archive, and through the persons corresponding to the faces in the face archive, the corresponding relationship between the person archive and the face archive is established, so as to realize the archiving of the body data, thereby judging the occurrence event of each person in the target area.
In the above scheme, the human body data in the face archive may not be accurate enough, and the human body data is recalled by establishing the corresponding relationship between the human body archive and the face archive only through the human body corresponding to the face in the face archive, so that part of the human body data is difficult to be correctly identified, which may result in a low recall rate of the human body data.
Disclosure of Invention
The application provides a human body data filing method, a human body data filing device, human body data filing equipment and a storage medium, and the recall rate of human body data filing is improved.
In one aspect, a human body data archiving method is provided, and the method includes:
acquiring at least two target images;
performing face clustering operation on the at least two target images to obtain face aggregation files;
classifying first human body data meeting a first similar condition with the associated human body data of the face aggregation files in the target image into the face aggregation files aiming at each face aggregation file to obtain each updated face aggregation file; the associated human body data is human body data associated with the human face data of the human face aggregation file;
for each updated face aggregation file, when human body data meeting a second similar condition with second human body data in the target image is searched in the updated face aggregation file, classifying the second human body data into the updated face aggregation file; the second human body data is human body data in the target image except the first human body data.
In yet another aspect, there is provided a human body data filing apparatus, the apparatus including:
the target image acquisition module is used for acquiring at least two target images;
the face clustering module is used for carrying out face clustering operation on the at least two target images to obtain each face aggregation file;
the first classification module is used for classifying first human body data meeting a first similar condition with the associated human body data in the target image into each human face aggregation file; the associated human body data are human body data associated with the human face data of each human face aggregation file;
and the second classification module is used for classifying the second human body data into the human face aggregation files corresponding to the third human body data when the third human body data meeting second similar conditions with the second human body data in the target image is searched in each human face aggregation file.
In still another aspect, a computer device is provided, where the computer device includes a processor and a memory, where the memory stores at least one instruction, at least one program, a code set, or a set of instructions, and the at least one instruction, at least one program, a code set, or a set of instructions is loaded and executed by the processor to implement the human body data archiving method.
In still another aspect, a computer-readable storage medium is provided, in which at least one instruction is stored, and the at least one instruction is loaded and executed by a processor to implement the human body data archiving method.
In yet another aspect, a computer program product is provided, as well as a computer program product or a computer program comprising computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device executes the human body data archiving method.
The technical scheme provided by the application can comprise the following beneficial effects:
when human body data in a video image needs to be filed, human faces can be aggregated firstly to obtain human face aggregation files corresponding to character information, and at the moment, in order to further judge action tracks of tasks corresponding to the human faces, the human body data in a target image can be compared with related human body data related to the human faces, so that the first human body data meeting similar conditions are further classified into the human face aggregation files; at the moment, part of the human body data which are related to the human face and do not meet similarity still exist in the target image, the computer equipment compares the similarity of the remaining second human body data with the human body data which are filed into each human face aggregation file, so that secondary judgment is carried out on the remaining human body data, incomplete classification caused by poor data quality of the human body data related to the human face is avoided as far as possible, and the recall rate of filing the human body data into the human face aggregation file is improved.
Drawings
In order to more clearly illustrate the detailed description of the present application or the technical solutions in the prior art, the drawings needed to be used in the detailed description of the present application or the prior art description will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a schematic structural diagram illustrating a person data archiving system according to an exemplary embodiment.
FIG. 2 is a method flow diagram illustrating a human data archiving method according to an exemplary embodiment.
FIG. 3 is a method flow diagram illustrating a human data archiving method according to an exemplary embodiment.
Fig. 4 is a diagram illustrating a logical framework of a person data archiving method according to an embodiment of the present application.
Fig. 5 is a block diagram illustrating a structure of a human body data filing apparatus according to an exemplary embodiment.
FIG. 6 is a schematic diagram of a computer device provided in accordance with an exemplary embodiment of the present application.
Detailed Description
The technical solutions of the present application will be described clearly and completely with reference to the accompanying drawings, and it should be understood that the described embodiments are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
It should be understood that "indication" mentioned in the embodiments of the present application may be a direct indication, an indirect indication, or an indication of an association relationship. For example, a indicates B, which may mean that a directly indicates B, e.g., B may be obtained by a; it may also mean that a indicates B indirectly, for example, a indicates C, and B may be obtained by C; it can also mean that there is an association between a and B.
In the description of the embodiments of the present application, the term "correspond" may indicate that there is a direct correspondence or an indirect correspondence between the two, may also indicate that there is an association between the two, and may also indicate and be indicated, configure and configured, and so on.
In the embodiment of the present application, "predefining" may be implemented by saving a corresponding code, table, or other manners that may be used to indicate related information in advance in a device (for example, including a terminal device and a network device), and the present application is not limited to a specific implementation manner thereof.
FIG. 1 is a block diagram of a people data archiving system according to an exemplary embodiment. The person data filing system includes a server 110 and a terminal 120. The terminal 120 and the server 110 perform data communication via a communication network, which may be a wired network or a wireless network.
Optionally, an application having an image processing function is installed in the terminal 120, and the application may be a professional image processing application, a social contact application, a virtual reality application, or an AI application having an image processing function, which is not limited in this embodiment of the present application.
Optionally, the terminal 120 may be a terminal device having an image capturing component, where the image capturing component is used to obtain an image and store the image in a data storage module in the terminal 120; the terminal 120 can also be a terminal device having a data transmission interface for receiving image data captured by an image capture device (e.g., a camera) having an image capture component.
Optionally, the terminal 120 may be a mobile terminal such as a smart phone, a tablet computer, a laptop portable notebook computer, or the like, or a terminal such as a desktop computer, a projection computer, or the like, or an intelligent terminal having a data processing component, which is not limited in this embodiment of the application.
The server 110 may be implemented as one server, or may be implemented as a server cluster formed by a group of servers, which may be physical servers or cloud servers. In one possible implementation, the server 110 is a backend server for applications in the terminal 120.
In a possible implementation manner of the embodiment of the application, each target image of the target area acquired by the image acquisition device is stored in the server 110, and when the action trajectory of each person in the target area needs to be analyzed, the target image may be subjected to face data archiving (i.e., classifying the face data into a face aggregation file) first, then human body data archiving (i.e., classifying the human body data into a human body data file) is performed, a face aggregation file and a human body aggregation file are obtained respectively, and then the relationship between the human body aggregation file and the face aggregation file is obtained, so that the person information (i.e., the human body information and the face information) in the target image is associated with each person identity, and the action trajectory of each person in the target area is determined according to the position and time of the face data and the human body data.
Optionally, the server may be an independent physical server, a server cluster formed by multiple physical servers, or a distributed system, and may also be a cloud server that provides technical operation and computation services such as cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a network service, cloud communication, middleware service, a domain name service, a security service, a CDN, a big data and artificial intelligence platform, and the like.
Optionally, the system may further include a management device, where the management device is configured to manage the system (e.g., manage connection states between the modules and the server, and the management device is connected to the server through a communication network. Optionally, the communication network is a wired network or a wireless network.
Optionally, the wireless network or wired network described above uses standard communication techniques and/or protocols. The network is typically the internet, but may be any other network including, but not limited to, a local area network, a metropolitan area network, a wide area network, a mobile, a limited or wireless network, a private network, or any combination of virtual private networks. In some embodiments, data exchanged over the network is represented using techniques and/or formats including hypertext markup language, extensible markup language, and the like. All or some of the links may also be encrypted using conventional encryption techniques such as secure sockets layer, transport layer security, virtual private network, internet protocol security, and the like. In other embodiments, custom and/or dedicated data communication techniques may also be used in place of, or in addition to, the data communication techniques described above.
FIG. 2 is a method flow diagram illustrating a human data archiving method according to an exemplary embodiment. The method is executed by a computer device, which may be a server or a terminal in a character data archiving system as shown in fig. 1. As shown in fig. 2, the human body data archiving method may include the following steps:
Alternatively, the at least two target images may be images acquired by the same image acquisition device at different times.
Alternatively, the at least two target images may be images acquired by different image acquisition devices at different times. In this case, the different image capturing devices may be image capturing devices deployed in the same area, such as cameras. That is, the target image at this time indicates image information of different positions of a certain area.
When the character information exists in the target images, the at least two target images can reflect the action track of a certain character in the target area. For example, if the target image a indicates that the person a appears at a first time and the target image B indicates that the person a appears at a second time, it is possible to determine, from the target image a and the target image B, that the person a has moved from the position of the target image a to the position of the target image B between the first time and the second time.
After the target images are obtained, face clustering operation is performed on at least two target images to cluster the marked face data in the target images into face aggregation files, and at this time, the face data with the same characteristics are clustered in the face clustering files, that is, the face clustering files can be regarded as a set of face data corresponding to different character information.
The associated human body data is human body data associated with the face data of each face aggregation file.
In the embodiment of the present application, the first human body data is any one of the human body data acquired by the target image.
When each face cluster file is acquired, each face data exists in the face cluster file, for example, in a certain face cluster file, each face data exists in the face cluster file, and each face data theoretically corresponds to the face data of the same person, for example, the face information ID is determined as the person a. However, the target image is usually acquired by simultaneously acquiring a human face and a human body, and therefore when the human face data corresponding to the person a is determined, the human body data connected with the human face data corresponding to the person a in the target image can be used as the associated human body data of the person a to represent the human body characteristics of the person a. The computer device can classify the related human body data of the human body A into the human face aggregation file as the human body data of the person indicated by the human face aggregation file.
Because each human body data also exists in the target image, in order to judge the character information represented by the human body data, the computer equipment can compare the human body data with the associated human body data in the face clustering file, and when the similarity between the human body data and the associated human body data is high, the human body data and the associated human body data can be considered as the human body data acquired for the same character.
Therefore, for each face aggregation file, the computer device can compare the similarity of the associated human body data in the face aggregation file with the similarity of each human body data in the target image, and when a first similarity condition is met, the similarity of the human body data and the associated human body data is higher, so that the human body data can be classified into the face aggregation file where the associated human body data is located.
For example, for the face aggregation file corresponding to the person a, when it is detected that certain body data and the associated body data in the face aggregation file (i.e., the body data connected to the face of the person a) satisfy the first similarity condition, the body data is likely to be the body data of the person a, and therefore the body data is classified as the first body data into the face aggregation file of the person a.
The second human body data is the human body data except the first human body data in the target image.
When first human body data corresponding to each person is detected in the target image through each piece of associated human body data, and the detected first human body data is filed in each face aggregation file to generate an updated face aggregation file, there may still exist residual human body data, namely second human body data, in the target image, which do not satisfy the first similarity condition with each piece of associated human body data. However, the second human body data which does not satisfy the first similar condition with the associated human body data does not represent the human body data of the person who must not aggregate the files for each face, and may not satisfy the first similar condition due to poor image quality of the associated human body data corresponding to the face data; or due to the problem of the acquisition angle of the image acquisition equipment, the two human body data of the same person are acquired from different human body orientations, so that the associated human body data and the second human body data are theoretically the data of the same human body, but the calculation process does not conform to the first similar condition.
Therefore, in order to avoid erroneous judgment of the human body data caused by the above situation, the computer device may perform secondary filing on the remaining human body data, and perform similarity calculation on each human body data through the first human body data and the associated human body data which have been filed in the face aggregation file.
For each updated face aggregation file, when human body data satisfying a second similarity condition with second human body data in the target image is searched in a certain face aggregation file, the human body data can be regarded as the same person information as the second human body data, and therefore the second human body data is classified into the face aggregation file.
In summary, when human body data in a video image needs to be archived, human faces may be aggregated first to obtain a human face aggregation file corresponding to character information, and at this time, in order to further determine an action track of a task corresponding to each human face, human body data in a target image may be compared with associated human body data related to the human face, so that first human body data meeting similar conditions is further categorized into each human face aggregation file; at the moment, partial human body data which are not similar to each other and are related to the human face still exist in the target image, the computer equipment compares the similarity of the residual second human body data with the human body data which are filed in each human face aggregation file, and therefore secondary judgment is conducted on the residual human body data, incomplete classification caused by poor data quality of the human body data related to the human face is avoided as far as possible, and the recall rate of filing the human body data into the human face aggregation files is improved.
FIG. 3 is a method flow diagram illustrating a human data archiving method according to an exemplary embodiment. The method is executed by a computer device, which may be a server or a terminal in the person data archiving method system as shown in fig. 1. As shown in fig. 3, the human body data archiving method may include the steps of:
In a possible implementation manner of the embodiment of the present application, the target image is image information subjected to data preprocessing, and a mark for a human face region and a mark for a human body region exist in the target image. That is, after the computer device acquires the target image, the data of the face image and the data of the human body image in the target image may also be acquired according to the mark of the face region and the mark of the human body region.
In a possible implementation manner, the target image may be an image processed by a target area detection algorithm, and after the target image is detected by the target area detection algorithm, each face image and each body image in the target image may be captured, that is, a detection frame of the face area and a detection frame of the body area are generated, so as to mark the face area and the body area.
After the target images are obtained, clustering operation can be performed on the face data in each target image, so that the face data with higher similarity in each face data is aggregated into a face aggregation file.
Optionally, the computer device obtains the target images one by one, determines whether a high-quality face exists in the target image (if the image quality of the face data is greater than the first quality threshold), and further determines the high-quality face when the high-quality face exists in the image.
And the computer equipment performs clustering operation on the high-quality face and the face data in each existing face aggregation file, and classifies the high-quality face into the face aggregation file when the high-quality face and the existing face aggregation file meet a matching condition (if the similarity is greater than a threshold value).
If the face aggregation file meeting the matching condition with the high-quality face is not detected, judging whether the high-quality face meets the filing requirement (if the image quality of the face data is greater than a second quality threshold), creating a face aggregation file, and classifying the high-quality face into the newly created face aggregation file.
In the embodiment of the application, after the clustering operation of the face data corresponding to the face image is completed, a human body clustering operation may be performed on each target image, for example, after each human body data in the target image is acquired, each human body data is clustered, so as to obtain a human body aggregate file.
In a possible implementation manner, when a target image is acquired, whether a high-quality face exists in the target image is detected, and when the high-quality face exists, face clustering operation is performed on the target image; and when no high-quality face exists, judging whether human body data exist in the target image, and when the human body data exist, executing human body clustering operation on the human body data.
In one possible implementation manner, when target human body data exists in a target image, human body data similar to the target human body data is searched within a specified time period of the target acquisition equipment; and when human body data which are acquired by the target acquisition equipment within a specified time period and are similar to the target human body data exist, the human body data are considered as valid data, and human body clustering operation is performed on the human body data.
And when human body data which are acquired by the target acquisition equipment within a specified time period and are similar to the target human body do not exist, acquiring the human body data which are acquired by adjacent equipment (such as equipment belonging to the same region as the target acquisition equipment) of the target acquisition equipment within the specified time period, and when human body data which are similar to the target human body data exist in the human body data acquired by the adjacent equipment within the specified time period, executing human body clustering operation on the human body data.
And when human body data which are acquired by adjacent equipment in a specified time period and are similar to the target human body data do not exist, the human body data can be considered as invalid human body data, and clustering processing is not performed on the human body data.
In the embodiment of the application, the computer equipment firstly carries out face clustering operation on each face data in the target image to obtain a face aggregation file; and then carrying out human body clustering operation on each human body data in the target image to obtain a human body aggregation file. At this time, in order to associate the human body data in the human body aggregation files with the face data so as to determine the identity information corresponding to each human body aggregation file, the computer device may perform similarity determination on the human body data in the human body aggregation files, and classify the first human body data satisfying the first similarity condition with the associated human body data into the corresponding face aggregation files.
In a possible implementation manner, based on the similarity between the associated human body data and each human body data in each human body aggregation file, the file similarity between the human face aggregation file and each human body aggregation file is obtained, and the human body aggregation file with the file similarity larger than a similarity threshold is determined as a target human body aggregation file;
classifying first human body data in the target human body aggregation file, wherein the spatiotemporal information between the first human body data and the associated human body data meets a first spatiotemporal relationship, into the human face aggregation file; the spatiotemporal information is used to indicate the time and location of the person's presence.
That is to say, when each face aggregation file is targeted, taking one face aggregation file as an example, when the computer device performs filing operation, the computer device compares the similarity of the associated human body data in the face aggregation file with each human body data in each human body aggregation file, at this time, the overall situation of the similarity of the associated human body data and each human body aggregation file can be determined, and thus the file similarity of the associated human body data and each human body aggregation file is obtained.
In a possible implementation manner, the associated human body data and the mean value of the vector distances of the individual human body data in the target human body aggregation file are obtained as the file similarity of the associated human body data and the target human body aggregation file.
In the process of calculating the similarity of the profiles, taking any one of the human body aggregation profiles as an example (such as the first human body aggregation profile), because the human body data existing in the first human body aggregation file is obtained by the clustering algorithm, therefore, each human body data in the first human body aggregation file can be preliminarily considered as the human body information of the same person, at the moment, each human body data in the first human body aggregation file is compared with each associated human body data in a similarity manner, namely, each human body data in the first human body aggregation file and the associated human body data can be respectively solved for the vector distance, and obtaining the mean value of the vector distances between the associated human body data and each human body data, wherein the mean value of the vector distances is the similarity of each candidate human body data in the first human body aggregation file.
Therefore, in a possible implementation manner, when the mean value of the vector distances between the related human body data and each human body data in the first human body aggregation file is greater than the similarity threshold, it is obvious that the personal information indicated by the first human body aggregation file is the same as the personal information indicated by the human face aggregation file with a greater probability, and the computer device may determine the first human body aggregation file as the target human body aggregation file.
When the associated human body data in the human face aggregation file and the first human body aggregation file are subjected to similarity calculation aiming at each human face aggregation file, and the file similarity of the associated human body data and each human body aggregation file is obtained, the computer equipment can determine the first human body aggregation file with the file similarity larger than the similarity threshold value as a target human body aggregation file, which represents that the target human body aggregation file is possibly the same as the person indicated by the human face aggregation file.
However, in the process of aggregating the human body data, due to errors of an aggregation algorithm, the human body data corresponding to different people may be aggregated in the same human body aggregation file, and in order to avoid filing the different people into the human face aggregation file as much as possible, the computer device may also screen the human body data according to a spatiotemporal relationship.
The computer device can acquire each human body data in the target human body aggregation file, acquire spatiotemporal information therein, namely the time when the image device acquires each human body data and the position of the image device acquiring the human body data, and screen the human body data according to the spatiotemporal relationship.
When the computer device filters the human body data, taking the first human body data as an example, obtaining spatiotemporal information of the first human body data (namely, time information and position information of a target image corresponding to the first human body data, namely, time and position of a person captured), and spatiotemporal information of associated human body data (namely, time information and position information of a target image corresponding to the second human body data, namely, time and position of the person captured), when the spatiotemporal information of the first human body data and the spatiotemporal information of the associated human body data do not satisfy a first spatiotemporal relationship (if a ratio of a position difference to the time difference is greater than a threshold), the first human body data and the associated human body data have a low possibility of corresponding to the same person information; correspondingly, when the first time-space relationship is satisfied, the probability that the first person data and the associated person data correspond to the same person information is higher, and at this time, the first person data can be classified into the face aggregation file.
After the computer equipment classifies the first human body data into the face aggregation file, part of candidate human body data which do not satisfy the first spatiotemporal relationship exist in the human body aggregation file, and in a possible implementation mode, the computer equipment judges whether spatiotemporal information between the candidate human body data and the associated human body data satisfies the second spatiotemporal relationship; the candidate human body data are the human body data of which the spatio-temporal information between the target human body aggregation file and the associated human body data does not satisfy a first spatio-temporal relationship;
when the candidate human body data and the spatio-temporal information before the associated human body data meet a second spatio-temporal relationship, adding the candidate human body data into an alternative set;
the computer equipment acquires a target action track formed by all human body data in the face aggregation file;
and when the position information of the third human body data in the alternative set is matched with the target action track, classifying the third human body data into the face aggregation file.
When the spatiotemporal information of the human body data in the human body aggregation file and the spatiotemporal information of the first associated human body data do not satisfy the first spatiotemporal relationship, that is, the candidate human body data classified into the face aggregation file are not present, the computer device may further determine the spatiotemporal information, that is, determine whether the candidate human body data and the associated human body data satisfy the second spatiotemporal relationship, where the second spatiotemporal relationship is slightly lower than the requirement of the first spatiotemporal relationship, for example, the ratio of the allowed position to the time may be slightly higher than the first spatiotemporal relationship, when the second spatiotemporal relationship is satisfied, the person still has a possibility of moving between the position corresponding to the candidate human body data and the position corresponding to the associated human body data, and then add the candidate human body data into the alternative set.
And the computer equipment judges each candidate human body data in the alternative set, directly acquires the position information of each human body data in the face aggregation file at the moment, constructs a target action track, and represents that the third human body data is probably the same person corresponding to the associated human body data when detecting that the position information of the third human body data in the candidate human body data meets the target action track, so that the third human body data is classified into the face aggregation file.
At this time, the human body aggregation file is classified at least once through the computer device, and the human body data similar to the associated human body data in the human face aggregation file is classified into the human face aggregation file, so that the remaining human body data in each human body aggregation file is the human body data which does not meet the recall condition.
However, in order to increase the recall rate of the human body aggregation files, the computer device may perform a secondary classification process on each human body aggregation file.
When the face aggregation file after the first classification (i.e., the updated face aggregation file) contains the face data of the corresponding person, the face aggregation file also contains the body data of the corresponding person obtained by the first classification. Therefore, when the candidate human body data of the human body aggregation files need to be classified for the second time, the computer device can judge the remaining human body data in each human body aggregation file again according to the human body data stored in the updated human body aggregation files for each human face aggregation file.
In a possible implementation manner of the embodiment of the application, for the remaining human body data in each human body aggregation file, that is, the second human body data in the target image, the computer device may compare the similarity of the second human body data with the human body data in each updated human face aggregation file, and aggregate the human face aggregation files corresponding to the human body data that satisfy the second similarity condition.
In a possible implementation manner of the embodiment of the application, for each updated face aggregation file, the computer device searches, as second human body data, human body data with the highest sum of similarities (or average of similarities) with each human body data of the face aggregation file among the remaining human body data, where the second human body data obviously has the highest similarity with the face aggregation file, and thus the computer device may obtain spatiotemporal information of the second human body data, so as to perform re-judgment on the second human body data from a spatiotemporal perspective.
Optionally, the second similarity condition may also be that the vector distance between the human body data is smaller than a threshold.
When it is detected that the human body data in the face aggregation file and the second human body data meet the second similarity condition, it is indicated that the human body data in the face aggregation file has the person information with a certain probability that is the same as that of the second human body data, that is, the human body data in the face aggregation file and the second human body data may be the collected human body data of the same person.
The computer device can acquire the spatiotemporal information of the second human body data and judge the spatiotemporal information so as to further determine the relationship between the second human body data and the person indicated by the face aggregation file.
In a possible implementation manner, the computer device acquires the second human body data in the alternative set; the second human body data is data except the third human body data in the alternative set;
and searching the human body data meeting a second similar condition with the second human body data in each updated human face aggregation file aiming at each second human body data in the alternative set.
After the computer equipment carries out a first filing process through the first similarity and the first time-space relationship, the remaining human body data meeting the second time-space relationship in the human body aggregated data are input into the alternative set, and then the human body data are subjected to supplementary filing through the target action track, and at the moment, only the human body data meeting the second time-space relationship but not meeting the target action track remain in the alternative set.
The computer equipment can screen the remaining human body data in the alternative set, compare the similarity of each remaining human body data in the alternative set with the updated human face aggregation file, screen the human body data meeting a second similar condition with the second human body data in the updated human face aggregation file, and accordingly start a secondary filing process.
In a possible implementation manner, the second human body data is human body data in an image acquired by an image acquisition device at a specified position in the target region.
Optionally, the image capturing device at the designated position of the target area may be an image capturing device at an entrance. The second human body data is data containing the person access information, the identity of the person corresponding to the second human body data is confirmed, and the second human body data has significance for monitoring the person flowing in the target area, so that the secondary filing can only detect the second human body data collected by the image collecting device at the specified position of the target area, the accuracy of monitoring the person access information is guaranteed as far as possible, and meanwhile, the resource consumption of computer equipment is reduced.
When the spatiotemporal information of each human body data in the updated human face aggregate file and the spatiotemporal information of the second human body data both satisfy a third spatiotemporal relationship (for example, the ratio of the position difference to the time difference is smaller than a threshold), it indicates that the second human body data has a greater matching degree with the human body data in the updated human body aggregate file, whether the human body data is human body characteristics or the spatiotemporal relationship. Therefore, the second human body data is more likely to be the same as the human body data in the updated human face aggregation file, and the computer device classifies the second human body data into the updated human face aggregation file, so that the secondary filing operation of the second human body data is completed.
FIG. 4 is a diagram illustrating a logical framework of a person data archiving method according to an embodiment of the present application. As shown in fig. 4, the method for archiving personal data may include the following steps:
the business module pushes the human face and human body structural data (namely human face data and human body data in the target image) and event information (such as an in-out event) to the filing module, and the filing module in the computer equipment judges which strategy is adopted according to the limiting conditions such as data quality and the like.
And if the high-quality face exists, entering a face document gathering process, and if the matching rule is met, taking the corresponding PID as a document ID.
If the human body data do not exist, judging whether the human body data exist, if the human body data do not exist, carrying out short-time reconnection on the human body, searching similar human bodies in a single mirror (namely single image acquisition equipment) in a short time, if the human body data exist, archiving, if the human body data do not exist, searching neighborhood short-time cross-border human bodies (namely the human body data acquired by adjacent image acquisition equipment), and if the human body data do not exist, archiving.
Triggering the human body of the face in a time window at regular time to carry out trajectory reconnection, taking the human face and human body accumulation results in a specified time window, traversing the files aggregated by the human face in the time window, traversing the files, associating the human body with the human body (namely associating human body data), searching similar human bodies, traversing the similar human bodies, and if the similar human bodies meet the time-space limit matching condition (namely a first time-space relation), then the short-time reconnection trajectory (namely the trajectory formed by the human body data meeting the first similarity relation and meeting the first time-space relation) associated with the human body is put under the file, if not, the matching condition (namely the second time-space relation) of low limit is judged to be met, if so, the human body is added into the alternative set, after the traversal is completed, through the file face and the path track (namely the target action track) strictly matched with the human body, the alternative human body which accords with the path route and has reasonable space-time relationship is classified into the file.
After the human face reconnection is completed, the human body track which is not included in the human face file and carries in or out an event (namely the track formed by the human body data which does not satisfy the first time-space relationship, satisfies the first time-space relationship and is not matched with the target action track) in the human face reconnection is taken for secondary filing, the human body track which is not filed and carries the event is traversed, a similar human body is searched in the filed human body track, and if the human body has the similar human body and meets the time-space limit (namely meets the third time-space relationship), the human body is included in the file.
In the embodiment of the application, the data with the events are recalled for the second time, and the event tracks which cannot be filed in the first filing are recalled for the second time in a matching manner, so that the event recall rate is effectively improved.
In summary, when human body data in a video image needs to be archived, human faces may be aggregated first to obtain a human face aggregation file corresponding to character information, and at this time, in order to further determine an action track of a task corresponding to each human face, human body data in a target image may be compared with associated human body data related to the human face, so that first human body data meeting similar conditions is further categorized into each human face aggregation file; at the moment, part of the human body data which are related to the human face and do not meet similarity still exist in the target image, the computer equipment compares the similarity of the remaining second human body data with the human body data which are filed into each human face aggregation file, so that secondary judgment is carried out on the remaining human body data, incomplete classification caused by poor data quality of the human body data related to the human face is avoided as far as possible, and the recall rate of filing the human body data into the human face aggregation file is improved.
Fig. 5 is a block diagram illustrating a structure of a human body data filing apparatus according to an exemplary embodiment. The device comprises:
a target image obtaining module 501, configured to obtain at least two target images;
a face clustering module 502, configured to perform face clustering operation on the at least two target images to obtain face aggregation files;
a first classification module 503, configured to classify first human body data, which satisfies a first similarity condition with the associated human body data, in the target image into each face aggregation file; the associated human body data are human body data associated with the human face data of each human face aggregation file;
a second classifying module 504, configured to, when third human body data meeting a second similarity condition with the second human body data in the target image is searched in each face aggregation file, classify the second human body data into a face aggregation file corresponding to the third human body data.
In one possible implementation, the apparatus further includes:
the human body clustering module is used for performing human body clustering operation on the at least two target images to obtain each human body aggregation file;
the first classification module is further configured to,
classifying first human body data, which satisfies a first similar condition with the associated human body data of the human face aggregation files, in each human body aggregation file into the human face aggregation file;
in a possible implementation manner, the first classifying module further includes:
a first similarity obtaining unit, configured to obtain, based on the associated human body data, a similarity between the human face aggregated file and each human body aggregated file, and determine, as a target human body aggregated file, a human body aggregated file with a file similarity greater than a similarity threshold;
the first classification unit is used for classifying first human body data, of the target human body aggregate file, of which the spatiotemporal information with the associated human body data meets a first spatiotemporal relationship into the human face aggregate file; the spatiotemporal information is used to indicate the time and location of the person's presence.
In a possible implementation, the first classification unit is further configured to,
judging whether the spatiotemporal information between the candidate human body data and the associated human body data meets a second spatiotemporal relationship; the candidate human body data are human body data of which the spatio-temporal information between the target human body aggregation file and the associated human body data does not satisfy a first spatio-temporal relationship;
when the candidate human body data and the spatiotemporal information before the associated human body data meet a second spatiotemporal relationship, adding the candidate human body data into an alternative set;
acquiring a target action track formed by each human body data in the face aggregation file;
and when the position information of the third human body data in the alternative set is matched with the target action track, classifying the third human body data into the face aggregation file.
In one possible implementation, the apparatus further includes:
a data alternative module, configured to obtain the second human body data from the alternative set; the second human body data is data except for third human body data in the alternative set;
and the similar data searching module is used for searching the human body data meeting a second similar condition with the second human body data in each updated face aggregation file aiming at each second human body data in the alternative set.
In a possible implementation manner, the second human body data is human body data in an image acquired by an image acquisition device at a specified position in the target region.
In one possible implementation manner, the second classification module is further configured to,
acquiring space-time information of the second human body data;
and when the spatiotemporal information of the second human body data and the spatiotemporal information of each human body data in the updated face aggregation file meet a third spatiotemporal relationship, classifying the second human body data into the updated face aggregation file.
In summary, when human body data in a video image needs to be archived, faces may be aggregated first to obtain a face aggregation file corresponding to character information, and at this time, in order to further determine an action track of a task corresponding to each face, human body data in a target image may be compared with associated human body data related to the face, so that the first human body data meeting similar conditions is further categorized into each face aggregation file; at the moment, part of the human body data which are related to the human face and do not meet similarity still exist in the target image, the computer equipment compares the similarity of the remaining second human body data with the human body data which are filed into each human face aggregation file, so that secondary judgment is carried out on the remaining human body data, incomplete classification caused by poor data quality of the human body data related to the human face is avoided as far as possible, and the recall rate of filing the human body data into the human face aggregation file is improved.
Refer to fig. 6, which is a schematic diagram of a computer device according to an exemplary embodiment of the present application, the computer device including a memory and a processor, the memory storing a computer program, and the computer program when executed by the processor implementing the method.
The processor may be a Central Processing Unit (CPU). The Processor may also be other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components, or a combination thereof.
The memory, which is a non-transitory computer readable storage medium, may be used to store non-transitory software programs, non-transitory computer executable programs, and modules, such as program instructions/modules corresponding to the methods of the embodiments of the present invention. The processor executes various functional applications and data processing of the processor by executing non-transitory software programs, instructions and modules stored in the memory, that is, the method in the above method embodiment is realized.
The memory may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created by the processor, and the like. Further, the memory may include high speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory optionally includes memory located remotely from the processor, and such remote memory may be coupled to the processor via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
In an exemplary embodiment, a computer readable storage medium is also provided for storing at least one computer program, which is loaded and executed by a processor to implement all or part of the steps of the above method. For example, the computer-readable storage medium may be a Read-Only Memory (ROM), a Random Access Memory (RAM), a Compact Disc Read-Only Memory (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, and the like.
Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the application being indicated by the following claims.
It will be understood that the present application is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the application is limited only by the appended claims.
Claims (10)
1. A human body data archiving method, characterized in that the method comprises:
acquiring at least two target images;
performing face clustering operation on the at least two target images to obtain face aggregation files;
classifying first human body data meeting a first similar condition with the associated human body data of the face aggregation files in the target image into the face aggregation files aiming at each face aggregation file to obtain each updated face aggregation file; the associated human body data is human body data associated with the human face data of the human face aggregation file;
for each updated face aggregation file, when human body data meeting a second similar condition with second human body data in the target image is searched in the updated face aggregation file, classifying the second human body data into the updated face aggregation file; the second human body data is human body data in the target image except the first human body data.
2. The method of claim 1, further comprising:
performing human body clustering operation on the at least two target images to obtain each human body aggregation file;
the classifying the first human body data meeting a first similarity condition with the associated human body data of the face aggregation file in the target image into the face aggregation file includes:
and classifying first human body data meeting a first similar condition with the associated human body data of the face aggregation files in each human body aggregation file into the face aggregation files aiming at each face aggregation file.
3. The method according to claim 2, wherein the classifying the first human body data satisfying the first similarity condition with the associated human body data of the face aggregation profile in each of the human body aggregation profiles into the face aggregation profile comprises:
based on the associated human body data, respectively obtaining the similarity of the human face aggregation file and each human body aggregation file according to the similarity of the associated human body data and each human body data in each human body aggregation file, and determining the human body aggregation file with the file similarity larger than a similarity threshold value as a target human body aggregation file;
classifying first human body data, of the target human body aggregate file, of which the spatio-temporal information with the associated human body data meets a first spatio-temporal relationship into the human face aggregate file; the spatiotemporal information is used to indicate the time and location of the person's presence.
4. The method according to claim 3, wherein the classifying the first human body data satisfying a first similarity condition with the associated human body data of the face aggregation profile in each of the human body aggregation profiles into the face aggregation profile further comprises:
judging whether the spatiotemporal information between the candidate human body data and the associated human body data meets a second spatiotemporal relationship; the candidate human body data are human body data of which the spatio-temporal information between the target human body aggregation file and the associated human body data does not meet a first spatio-temporal relationship;
when the candidate human body data and the spatiotemporal information before the associated human body data meet a second spatiotemporal relationship, adding the candidate human body data into an alternative set;
acquiring a target action track formed by each human body data in the face aggregation file;
and when the position information of the third human body data in the alternative set is matched with the target action track, classifying the third human body data into the face aggregation file.
5. The method according to claim 4, wherein before searching the updated face aggregation file for human data satisfying a second similarity condition with second human data in the target image and classifying the second human data into the updated face aggregation file, the method further comprises:
acquiring the second human body data in the alternative set; the second human body data is data except for third human body data in the alternative set;
and searching the human body data meeting a second similar condition with the second human body data in each updated face aggregation file aiming at each second human body data in the alternative set.
6. The method according to claim 5, wherein the second human body data is human body data in an image acquired by an image acquisition device at a specified position in the target region.
7. The method according to any one of claims 1 to 6, wherein the classifying the second human body data into the updated face aggregation file comprises:
acquiring space-time information of the second human body data;
and when the spatiotemporal information of the second human body data and the spatiotemporal information of each human body data in the updated face aggregation file meet a third spatiotemporal relationship, classifying the second human body data into the updated face aggregation file.
8. A human data archiving apparatus, the apparatus comprising:
the target image acquisition module is used for acquiring at least two target images;
the face clustering module is used for carrying out face clustering operation on the at least two target images to obtain each face aggregation file;
the first classification module is used for classifying first human body data meeting a first similar condition with the associated human body data in the target image into each human face aggregation file; the associated human body data are human body data associated with the human face data of each human face aggregation file;
and the second classification module is used for classifying the second human body data into the human face aggregation files corresponding to the third human body data when the third human body data meeting second similar conditions with the second human body data in the target image is searched in each human face aggregation file.
9. A computer device comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a set of codes, or a set of instructions, which is loaded and executed by the processor to implement the human data archiving method as claimed in any one of claims 1 to 7.
10. A computer-readable storage medium, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the human body data archiving method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111681912.6A CN114519879A (en) | 2021-12-30 | 2021-12-30 | Human body data archiving method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111681912.6A CN114519879A (en) | 2021-12-30 | 2021-12-30 | Human body data archiving method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114519879A true CN114519879A (en) | 2022-05-20 |
Family
ID=81597364
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111681912.6A Pending CN114519879A (en) | 2021-12-30 | 2021-12-30 | Human body data archiving method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114519879A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115017359A (en) * | 2022-05-27 | 2022-09-06 | 浙江大华技术股份有限公司 | Method and device for searching picture and electronic equipment |
| CN115880727A (en) * | 2023-03-01 | 2023-03-31 | 杭州海康威视数字技术股份有限公司 | Training method and device for human body recognition model |
-
2021
- 2021-12-30 CN CN202111681912.6A patent/CN114519879A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115017359A (en) * | 2022-05-27 | 2022-09-06 | 浙江大华技术股份有限公司 | Method and device for searching picture and electronic equipment |
| CN115880727A (en) * | 2023-03-01 | 2023-03-31 | 杭州海康威视数字技术股份有限公司 | Training method and device for human body recognition model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20220092881A1 (en) | Method and apparatus for behavior analysis, electronic apparatus, storage medium, and computer program | |
| US20210357624A1 (en) | Information processing method and device, and storage medium | |
| CN114519879A (en) | Human body data archiving method, device, equipment and storage medium | |
| CN109740573B (en) | Video analysis method, device, equipment and server | |
| TWI789128B (en) | Face recognition method, device, equipment and storage medium | |
| CN110941978A (en) | Face clustering method and device for unidentified personnel and storage medium | |
| CN113378616A (en) | Video analysis method, video analysis management method and related equipment | |
| WO2020047084A1 (en) | Methods and apparatus for reducing false positives in facial recognition | |
| TW202125332A (en) | Method and device for constructing target motion trajectory, and computer storage medium | |
| CN113673311A (en) | Traffic abnormal event detection method, equipment and computer storage medium | |
| CN110879986A (en) | Face recognition method, apparatus and computer-readable storage medium | |
| CN112347296A (en) | Person and case association analysis method and device based on face recognition | |
| CN112614102A (en) | Vehicle detection method, terminal and computer readable storage medium thereof | |
| CN111522974A (en) | Real-time filing method and device | |
| CN116244609A (en) | Passenger flow volume statistics method and device, computer equipment and storage medium | |
| CN111091047B (en) | Living body detection method and device, server and face recognition equipment | |
| CN111914649A (en) | Face recognition method and device, electronic equipment and storage medium | |
| CN113837006B (en) | Face recognition method and device, storage medium and electronic equipment | |
| CN111753601A (en) | Image processing method and device and storage medium | |
| CN117671440A (en) | Abnormal portrait file detection method and system | |
| CN113822110A (en) | Target detection method and device | |
| CN112487082A (en) | Biological feature recognition method and related equipment | |
| CN115719428A (en) | Face image clustering method, device, equipment and medium based on classification model | |
| KR20160078154A (en) | Customer information provision method, device and computer program | |
| CN111563479B (en) | Concurrent person weight removing method, partner analyzing method and device and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |