Population database sub-database method and device

Abstract

The present application provides a method and apparatus for sub - dividing a population database. In the technical solution of the present application, according to the probability of a person appearing in a first camera, a set of persons who are likely to appear in the first camera is determined, and the sub - division of the population database corresponding to the area is determined according to the sets of persons in each first camera. Since the probability of a person appearing can be continuously updated, it is possible to always keep the most active actual population in the area in the sub - divided population database. Compared with static sub - division, it can greatly reduce invalid personnel data. When performing the face 1:N comparison and real - name tagging process, it can improve the hit rate of the first comparison, greatly reduce the second comparison, and achieve the effects of less consumption of comparison resources and short time consumption.

Description

Technical Field

[0001] The present application relates to the field of population management, and more specifically, to a method and device for dividing a population database. Background Art

[0002] The city-level intelligent portrait is to label dynamic faces in real time based on the sub-library list of dynamic faces and real names, so as to confirm the identity information of dynamic faces. In the process of real-time labeling of dynamic faces, a 1:N comparison will be performed between the dynamic face features and the static face feature library. Among them, dynamic faces are faces collected in non-constrained scenes without user perception, such as the faces of passers-by collected by face capture cameras; static faces are faces collected in specific constrained scenes with user perception, such as faces collected by ID photos.

[0003] As the size of the static facial feature database (or population database) N continues to increase, the accuracy of 1:N matching will decrease, and the resources and time consumed by the matching will increase. The city-level static facial feature database is usually in the tens of millions. In order to reduce the size of the static facial feature database N, the traditional approach is to divide the static facial feature database according to the administrative divisions to which the static personnel belong, that is, the static database division method.

[0004] In the static database division method of the population database, the permanent population is classified according to the administrative division where the household registration is located; the floating population is classified according to the administrative division where the registered residence is located. The granularity of the classification can be defined according to the level of the administrative division. The domestic administrative divisions are divided into four levels: province, city, county (district), township and street.

[0005] However, due to the mobility of the population, the permanent population sometimes does not live and reside in their registered place of residence; for the floating population living in the jurisdiction, due to the lag in updating the floating population database data, it is possible that they are not in the sub-database of the population database. In other words, the static sub-database method has a high probability of causing the population who actually live and reside in the jurisdiction to not be included in the sub-database of the population database, while the population who do not live and reside in the jurisdiction is included in the sub-database of the population database. This will result in a low hit rate for a 1:N comparison based on the sub-database of the population database, and even if certain comparison steps and logic are adopted, for example, the first comparison is completed in the sub-database of the jurisdiction at this level, and the second comparison is completed in the sub-database of the superior jurisdiction, it will also bring problems such as large consumption of comparison resources and long comparison time. Summary of the invention

[0006] The present application provides a population database partitioning method and device, which can partition the population database more accurately, thereby improving the hit rate of a 1:N comparison match based on the population database partition.

[0007] In a first aspect, the present application provides a method for partitioning a population database. The method includes: determining N first cameras within a first area, where N is a positive integer; determining N personnel sets based on the N first cameras, wherein the probability that the personnel in the i-th personnel set among the N personnel sets appear in the i-th camera is greater than zero, and the i-th personnel set among the N personnel sets corresponds to the i-th camera, and the value of i is each value in [1, N]; determining a first population partition of the first area based on the N personnel sets; storing the first population partition, where the first population partition is used to determine the personnel identity information corresponding to the first face data collected by the N first cameras.

[0008] For example, the personnel whose appearance probability in the i-th camera is greater than a certain threshold are included in the personnel set corresponding to the i-th camera. This threshold can be zero or a specific value.

[0009] Optionally, in the embodiments of the present application, different time periods can correspond to different population partitions. For example, a day can be divided into 1440 / T time periods, where T is the duration of each time period, and the 1440 / T time periods can correspond to 1440 / T population partitions.

[0010] In the above technical solution, based on the personnel appearance probability of personnel appearing in the first camera, a personnel set that may appear in the first camera is determined, and the population database partition corresponding to this area is determined based on the personnel sets of each first camera. Since the personnel appearance probability can be continuously updated, it is possible to always keep the most active actual population in the area in the population database partition. Compared with static partitioning, it can greatly reduce invalid personnel data. When performing the face 1:N comparison and real-name tagging process, the hit rate of a single comparison can be improved, and the secondary comparison can be significantly reduced, achieving the effects of less consumption of comparison resources and short time consumption.

[0011] In addition, the first population partition only includes personnel whose appearance probability is greater than zero, that is to say, the first population partition only includes personnel who may appear in the first camera. In this way, the data volume in the first population partition can be reduced, and the resource consumption for comparison can be reduced.

[0012] In a possible implementation manner, before determining the first population partition of the first area, the method further includes: obtaining second face data captured by M second cameras, where the second cameras are the cameras where the personnel in the i-th personnel set were located before the regional migration; determining the probability that the personnel in the i-th personnel set appear in the i-th camera based on the second face data and the personnel migration probability, where the personnel migration probability is the probability that the personnel migrate from the second camera to the i-th first camera.

[0013] In the above technical solution, according to the real-time face data from the camera, the probability of a person appearing is updated. That is to say, the list of person IDs corresponding to each camera can be dynamically updated. In this way, according to the area where the camera is located, the sub-library of the population database corresponding to each area is determined, so that the most vivid actual active population in the area can always be maintained in the sub-library of the population database. Compared with the full-scale static database, the invalid person data is greatly reduced. When performing the face 1:N comparison and real-name tagging process, the hit rate of the first comparison can be improved, and the secondary comparison can be greatly reduced, achieving the effects of less consumption of comparison resources and short time consumption.

[0014] In a possible implementation manner, determining the probability that a person in the i-th person set appears in the i-th camera according to the second face data and the person migration probability includes: comparing the similarity between the second face data and the face data in the second sub-library of the population, obtaining the confidence level of the second face data, where the confidence level is used to indicate the confidence probability that the second face data and the face data in the second sub-library of the population belong to the same person, and the second sub-library of the population is the sub-library of the population corresponding to the second camera; obtaining the probability that a person in the i-th person set appears in the i-th camera according to the confidence level and the person migration probability.

[0015] Optionally, the probability of a person appearing can be obtained by multiplying the confidence level corresponding to a certain face data by the person migration probability corresponding to the person.

[0016] In a possible implementation manner, the method further includes: determining the probability that a person migrates from the second camera to the i-th first camera according to the historical spatio-temporal trajectory data of each person in the permanent population database and / or the floating population database in the first area within a preset time period.

[0017] In a possible implementation manner, the method further includes: determining the initial probability that a person in the i-th person set appears in the i-th first camera according to the permanent population database and / or the floating population database in the first area.

[0018] Second aspect, the present application provides a population database sub - database device, the device includes: a processing unit, configured to determine N first cameras in a first area, where N is a positive integer; the processing unit is further configured to determine N personnel sets according to the N first cameras, wherein the probability that the personnel in the i - th personnel set among the N personnel sets appear in the i - th camera is greater than zero, and the i - th personnel set among the N personnel sets corresponds to the i - th camera, and the value of i is each value in [1, N]; the processing unit is configured to determine a first population sub - database of the first area according to the N personnel sets; a storage unit, configured to store the first population sub - database, and the first population sub - database is used to determine the personnel identity information corresponding to the first face data collected by the N first cameras.

[0019] For example, the personnel whose appearance probability in the i - th camera is greater than a certain threshold are included in the personnel set corresponding to the i - th camera. This threshold can be zero or a specific value.

[0020] Optionally, in the embodiments of the present application, different time periods can correspond to different population sub - databases. For example, a day can be divided into 1440 / T time periods, where T is the duration of each time period, and the 1440 / T time periods can correspond to 1440 / T population sub - databases.

[0021] In the above technical solution, according to the personnel appearance probability of personnel appearing in the first camera, the personnel set that may appear in the first camera is determined, and the population database sub - database corresponding to this area is determined according to the personnel sets of each first camera. Since the personnel appearance probability can be continuously updated, it is possible to always keep the most active actual population in the area in the population database sub - database. Compared with the static sub - database, the invalid personnel data can be greatly reduced. When performing the face 1:N comparison and real - name tagging process, the hit rate of a single comparison can be improved, and the secondary comparison can be significantly reduced, achieving the effects of less consumption of comparison resources and short time consumption.

[0022] In addition, the first population sub - database only includes personnel with a personnel appearance probability greater than zero. That is to say, the first population sub - database only includes personnel who may appear in the first camera. In this way, the data volume in the first population sub - database can be reduced, and the resource consumption for comparison can be reduced.

[0023] In a possible implementation, the device further includes: an acquisition unit, configured to acquire second face data captured by M second cameras before determining the first population sub-library of the first area, where the second cameras are the cameras where the personnel in the i-th personnel set were located before area migration; the processing unit is further configured to determine the probability that the personnel in the i-th personnel set appear in the i-th camera according to the second face data and the personnel migration probability, where the personnel migration probability is the probability that the personnel migrate from the second camera to the i-th first camera.

[0024] In the above technical solution, the probability of personnel appearance is updated according to the real-time face data from the cameras. That is to say, the personnel ID list corresponding to each camera can be dynamically updated. In this way, according to the area where the camera is located, the population sub-library corresponding to each area is determined, and the most active actual mobile population in the area can always be maintained in the population sub-library. Compared with the full static library, a large amount of invalid personnel data is greatly reduced. When performing the face 1:N comparison and real-name tagging process, the hit rate of the first comparison can be improved, and the secondary comparison can be greatly reduced, achieving the effects of less consumption of comparison resources and short time consumption.

[0025] In a possible implementation, the processing unit is specifically configured to: compare the similarity between the second face data and the face data in the second population sub-library to perform comparison, and obtain the confidence level of the second face data, where the confidence level is used to indicate the confidence probability that the second face data and the face data in the second population sub-library belong to the same person, and the second population sub-library is the population sub-library corresponding to the second camera; according to the confidence level and the personnel migration probability, obtain the probability that the personnel in the i-th personnel set appear in the i-th camera.

[0026] Optionally, the probability of personnel appearance can be obtained by multiplying the confidence level corresponding to a certain face data by the personnel migration probability corresponding to the personnel.

[0027] In a possible implementation, the processing unit is further configured to: determine the probability that the personnel migrate from the second camera to the i-th first camera according to the historical spatio-temporal trajectory data of each person in the permanent population library and / or the floating population library of the first area within a preset time period.

[0028] In a possible implementation, the processing unit is further configured to: determine the initial probability that the personnel in the i-th personnel set appear in the i-th first camera according to the permanent population library and / or the floating population library of the first area.

[0029] In a third aspect, the present application provides a chip, which is connected to a memory and is configured to read and execute a software program stored in the memory to implement the method described in the first aspect or any implementation manner of the first aspect.

[0030] In a fourth aspect, the present application provides a population database sub-library device, including a memory for storing a program, and a processor for executing the program stored in the memory. When the program stored in the memory is executed, the processor is configured to execute the method in the first aspect and any possible implementation manner in the first aspect.

[0031] In a possible implementation manner, the population database sub-library device further includes a transceiver.

[0032] In a possible implementation manner, the population database sub-library device is a chip that can be applied to a network device.

[0033] In a possible implementation manner, the population database sub-library device is a server, a cloud host, or a container.

[0034] In a fifth aspect, the present application provides a computer program product, which includes computer instructions. When the computer instructions are executed, the method in the foregoing first aspect or any possible implementation manner of the first aspect is executed.

[0035] In a sixth aspect, the present application provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed, the method in the foregoing first aspect or any possible implementation manner of the first aspect is executed. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the static sub-library of the population database.

[0037] Figure 2 It is a comparison diagram of the population in the population sub-library obtained by static sub-library and the actual active population.

[0038] Figure 3 It is a schematic flowchart of the population database sub-library method provided by the embodiment of the present application.

[0039] Figure 4 It is a schematic flowchart of the method for accurate population sub-library of the embodiment of the present application.

[0040] Figure 5 It is a schematic diagram of the processing flow for determining the initial appearance probability of the population in the embodiment of the present application.

[0041] Figure 6 It is an example of the initial appearance probability of a person in the embodiment of the present application.

[0042] Figure 7 It is a schematic diagram of the processing flow for determining the spatio-temporal migration probability matrix of personnel in an embodiment of the present application.

[0043] Figure 8 It is an example of the personnel migration probability distribution matrix in an embodiment of the present application.

[0044] Figure 9 It is a schematic diagram of the processing flow for updating the appearance probability of personnel in an embodiment of the present application.

[0045] Figure 10 It is an example of updating the appearance probability of personnel in an embodiment of the present application.

[0046] Figure 11 It is a schematic diagram of the secondary population database sub-library in an embodiment of the present application.

[0047] Figure 12 It is a schematic flowchart of personnel comparison based on the population sub-library in an embodiment of the present application.

[0048] Figure 13 It is a schematic flowchart of a retrieval comparison in an embodiment of the present application.

[0049] Figure 14 It is an implementation form of the system in an embodiment of the present application.

[0050] Figure 15 It is a schematic structural diagram of the population database sub-library device in an embodiment of the present application.

[0051] Figure 16 It is a schematic structural diagram of the population database sub-library device provided by another embodiment of the present application. Detailed implementation manners

[0052] Next, the technical solutions in the present application will be described in conjunction with the accompanying drawings.

[0053] The technical solutions of the embodiments of the present application can be applied to various scenarios as long as the scenarios require sub-library of the population database. For example, urban-level intelligent portrait scenarios, public security population management scenarios, intelligent transportation scenarios, etc. Next, taking the urban-level intelligent portrait scenario as an example, the technical solutions of the present application will be described.

[0054] Urban-level intelligent portrait is to perform real-time tagging on dynamic faces according to the sub-library list of dynamic faces and real names, so as to implement the personnel identity information of dynamic faces. During the process of real-time tagging of dynamic faces, 1:N comparison will be performed between dynamic face features and static face feature libraries.

[0055] Among them, face 1:N comparison refers to the comparison of a face and a set (including N comparison objects), and the query obtains the face with high similarity to the specified face. Dynamic faces are faces collected in unconstrained scenes without user perception, such as faces of passers-by collected by face capture cameras. Static faces are faces collected in specific constrained scenes with user perception, such as faces collected by ID photos. Facial features are feature vectors mapped by mapping pixels in a face close-up image. A face close-up image is an image of a face close-up area that meets the face recognition pixel requirements deducted from a face scene image, commonly known as a small face image. A face scene image is a captured image that contains at least one face and human body element, commonly known as a large face image.

[0056] As the size of the static facial feature database (or population database) N continues to increase, the accuracy of 1:N matching will decrease, and the resources and time consumed by the matching will increase. The city-level population database is usually in the tens of millions. In order to reduce the size of the population database N, the traditional approach is to divide the population database according to the administrative divisions to which the static people belong, that is, the static database division method.

[0057] Figure 1 This is a schematic diagram of the static sub-database of the population database. Figure 1 As shown in the figure, in the static sub-database method of the population database, the permanent population is classified according to the administrative division where the household registration is located; the floating population is classified according to the administrative division where the registered residence is located. The granularity of the classification can be defined according to the level of the administrative division. The domestic administrative divisions are divided into four levels: province, city, county (district), township and street.

[0058] However, due to population mobility, permanent residents sometimes do not live and reside in their registered place of residence; for floating population living in the jurisdiction, due to the lag in updating the floating population database data, it is possible that they are not included in the population database sub-database. Figure 2 The static database partitioning method shown has a high probability of causing the population that actually lives and resides in the jurisdiction to not be included in the population database partition, while the population that does not live and reside in the jurisdiction is included in the population database partition, making the population database partition lack freshness. This will result in a low hit rate for a 1:N match based on the population database partition, and even if certain matching steps and logic are adopted, for example, the first match is completed in the database of the jurisdiction at this level, and the second match is completed in the database of the higher-level jurisdiction, it will also bring about problems such as large consumption of matching resources and long matching time.

[0059] In response to the above problems, the embodiments of the present application provide a population database partitioning method and device, which can partition the population database more accurately, thereby improving the hit rate of a 1:N comparison match based on the population database partitioning.

[0060] Figure 3 It is a schematic flowchart of the method for dividing a population database into sub-databases provided by an embodiment of the present application. Figure 3 The method shown can be executed by a server, a cloud host, a container, etc., or can be executed by a chip or module included in a server, a cloud host, a container, etc. Figure 3 The method shown includes at least some of the following content.

[0061] In 310, determine N first cameras in a first area, where N is a positive integer.

[0062] The first area can be an area of any size, and the embodiments of the present application do not make specific limitations. For example, it can be administrative divisions such as provinces, cities, counties, districts, streets, etc. Again, for example, it can be an area including a preset number of cameras.

[0063] The camera can also be other devices or equipment with a photographing function, and the embodiments of the present application do not make specific limitations.

[0064] The N first cameras in the first area are all the cameras deployed or set in the first area.

[0065] In a possible implementation, for each area, at least one camera list can be saved. When determining the N first cameras in the first area, the camera list corresponding to the first area can be determined according to the identity (ID) of the first area, etc. Among them, the camera list can be determined according to the geographical location of the camera and the ID of the camera, and the geographical location of the camera can refer to the longitude and latitude where the camera is located, etc.

[0066] In another possible implementation, the area where each first camera is located can be directly determined according to the geographical location of the first camera and the ID of the first camera, so as to determine the N first cameras in the first area. Among them, the geographical location of the camera can refer to the longitude and latitude where the camera is located, etc.

[0067] In 320, according to the N first cameras, determine N personnel sets, where the probability that the personnel in the i-th personnel set among the N personnel sets appear in the i-th camera is greater than zero (hereinafter referred to as the personnel appearance probability), and the i-th personnel set among the N personnel sets corresponds to the i-th camera, and the value of i is each value in [1, N].

[0068] In 330, according to the N personnel sets, determine the first population sub-database of the first area.

[0069] In 340, store the first population sub-database.

[0070] In a possible implementation, according to the probabilities of the people in the N sets of people appearing in the corresponding first cameras, the N sets of people corresponding to the N first cameras are determined, and further, according to the N sets of people, the population sub-library of the first area is determined. Among them, the probability of the people in the set of people appearing in the first camera is greater than 0. For example, cameras 1-3 are set in the first area. There are 300 people who may appear in camera 1, 60 people who may appear in camera 2, and 15 people who may appear in camera 3. Then the population sub-library of the first area consists of some or all of the 375 people.

[0071] In another possible implementation, the population sub-library in the embodiments of the present application can be divided by time slices. Specifically, according to the probabilities of the people in the N sets of people in the first time period appearing in the corresponding first cameras, the N sets of people corresponding to the N first cameras in the first time period are determined, and further, according to the N sets of people, the population sub-library of the first area in the first time period is determined; according to the probabilities of the people in the N sets of people in the second time period appearing in the corresponding first cameras, the N sets of people corresponding to the N first cameras in the second time period are determined, and further, according to the N sets of people, the population sub-library of the first area in the second time period is determined. That is to say, the first area can correspond to different population sub-libraries in different time periods. The time period in the embodiments of the present application can be a time slice divided according to a preset duration. For example, the length of the time slice is T, and 1440 minutes in a day can be divided into 1440 / T time slices, and T can be divisible by 1440. In this example, the time slice is divided in a cycle of days and in minutes as the unit. It can be understood that the time slice in the embodiments of the present application can also be divided in other granularity cycles and / or other granularity units.

[0072] Before executing 320, the probability of the people appearing as described above can also be determined first.

[0073] In a possible implementation, when the M second cameras capture the face data of the people in the set of people corresponding to the i-th first camera, the probability of the people in the i-th set of people appearing is triggered to be updated. Among them, the second camera is the camera where the people in the set of people corresponding to the first camera are located before the area migration. Specifically, when the M second cameras obtain the face data, according to the face data and the people migration probability, the probability of the people in the i-th set of people appearing in the i-th camera is determined, where the people migration probability is the probability of the people migrating from the second camera to the i-th first camera.

[0074] More specifically, as an example, the face data is compared in the second population sub-library to obtain the confidence level of the face data. Based on the obtained confidence level and the personnel migration probability, the probability that a person in the i-th personnel set appears in the i-th camera is obtained. Among them, the confidence level is used to indicate the confidence probability that the second face data and the face data in the second population sub-library belong to the same person. For example, it can be a similarity, etc.; the second population sub-library is the population sub-library corresponding to the second camera.

[0075] Optionally, the above-mentioned personnel migration probability can be applicable to each person in the first area.

[0076] Optionally, the above-mentioned personnel migration probability can also be specific to each person. For example, for person A, the probability that he migrates from the second camera to the first camera is zero, which means that person A will not migrate from the second camera to the first camera; for person B, the probability that he migrates from the second camera to the first camera is not zero, which means that person B may migrate from the second camera to the first camera.

[0077] It should also be understood that when the face data captured by the first camera is compared in the first population sub-library, the probability of migration of the personnel in the personnel set corresponding to the third camera to the third camera can also be updated. For the specific update method, refer to the update method in the first camera above, which will not be elaborated here.

[0078] Based on the N personnel sets, the first population sub-library of the first area is determined. In one possible implementation, the personnel with a non-zero appearance probability in the first camera can be included in the first population sub-library, that is, the personnel who appear in the second camera and have a non-zero probability of migrating from the second camera to the first camera can be included in the first population sub-library.

[0079] In another possible implementation, a real-time accurate population sub-library can be formed according to the personnel appearance probability, the area corresponding to the geographical location of the front-end camera, and the retrieval logic of personnel filing. Specifically, the secondary sub-library is divided by district and county, and the sub-district office; the primary sub-library is divided by city; the full library volume includes the full amount of permanent residents and floating population, generally by province or city.

[0080] Furthermore, each level of the sub-library can be divided into one or multiple levels of accurate sub-libraries according to the accuracy. Taking the two-level accurate sub-library as an example, the first accurate sub-library includes the personnel with a non-zero appearance probability, and the second accurate sub-library includes the personnel who have never appeared in the area (that is, the personnel appearance probability is 0) and the personnel with a non-zero appearance probability. In this way, when comparing personnel, the comparison can be preferentially carried out in the first accurate sub-library, further improving the hit rate of the first personnel 1:N comparison and reducing the secondary comparison.

[0081] Before executing 320, the migration probability of personnel can also be determined first.

[0082] Optionally, the probability of a person migrating from the second camera to the first camera can be determined according to the historical spatio-temporal trajectory data of each person in the set of N persons within a preset time period.

[0083] Optionally, the probability of a person migrating from the second camera to the first camera can be determined according to the historical spatio-temporal trajectory data of all or some of the persons in the permanent population database and / or floating population database of the first region within a preset time period.

[0084] Optionally, the preset time period can be any possible time length, for example, 1 month, 1 week, 3 months, 1 year, etc. Among them, the historical spatio-temporal trajectory data can be the activity trajectory data of each person every day, and this data can include the captured time, the camera that captured the image, etc.

[0085] It can be understood that when the system is initialized, since there is no real-time data of personnel, an initial population sub-database can be generated through administrative divisions and real-time population distribution data. Specifically, according to the permanent population database and / or floating population database of the first region, the initial probability of the personnel in the i-th set of personnel appearing in the i-th first camera is determined.

[0086] Next, in combination with specific examples, the technical solutions of the embodiments of the present application will be described in more detail.

[0087] Figure 4 It is a schematic flowchart of a method for accurately dividing the population of a sub-database that perceives the spatio-temporal law of personnel migration according to the embodiments of the present application.

[0088] As Figure 4 shown, in 410, after the system initialization is started, the data of the permanent population database and floating population database with real names in administrative divisions is imported, and based on the administrative divisions of the permanent population and floating population, and the geographical locations of each camera, the initial appearance probability of personnel and the initial population sub-database are generated.

[0089] Specifically, as Figure 5 shown, the input information of the system includes:

[0090] (1) Camera ID and the geographical location of the camera, for example, longitude and latitude, etc.

[0091] (2) Personnel information divided by provincial-city-county administrative regions, and the personnel information includes permanent population information and floating population information.

[0092] The processing flow of the initial appearance probability of personnel includes:

[0093] In 4101, when the system is initially launched, since there is no real-time data of personnel, an initial population sub-library is generated based on administrative divisions and real-time population distribution data, or in other words, an initial population sub-library for each administrative division is generated according to the permanent population library and the floating population library. The initial population sub-library records information such as the ID of the personnel, the facial data of the personnel, spatial trajectory data, and place of residence.

[0094] For example, if the registered place of residence of person A is administrative division B, then person A belongs to the population sub-library of administrative division B.

[0095] In 4102, based on the initial population sub-library obtained in 4101, the camera ID, and the geographical location of the camera, a list of personnel IDs for each camera is established. This list of personnel IDs includes the IDs of the personnel who may appear in this camera. The list of personnel IDs can correspond to the personnel set mentioned above.

[0096] For example, if the registered place of residence of person A is administrative division B, then person A is included in the list of personnel IDs of the camera whose geographical location is in administrative division B.

[0097] In addition to the place of residence, the corresponding relationship between the personnel ID and the camera can also be established using information such as the place of birth, the place of work, and the place of consumption.

[0098] In 4103, when the system is initially started, an initial appearance probability is assigned to the personnel IDs in the initial personnel sub-library. This appearance probability includes the initial appearance probability within each time slice. Optionally, the initial appearance probabilities of permanent personnel and floating personnel can be different. For example, the initial appearance probability of permanent personnel is p%, and that of floating personnel is q%. p and q can take any possible values, and the embodiments of the present application do not make specific limitations. For example, p = 5, q = 10; p = 15, q = 10; p = 25, q = 40; p = 50, q = 20; p = 5.5, q = 1.3, etc. Of course, the values of p and q can also be the same.

[0099] Figure 5 The probability values in the table shown in represent the probability that a certain person appears in the corresponding camera. For example, within the time slice (0:00, T), the appearance probability that the person corresponding to personnel ID1 appears in camera 1 is p%, and the appearance probability that the person corresponding to personnel ID3 appears in camera 1 is q%.

[0100] Among them, the time slice can be a period of time. For example, 1440 minutes in 1 day can be divided into 1440 / T time periods according to the duration T, and each time period can be regarded as a time slice, and the value of T is not limited.

[0101] The setting method of the initial appearance probability described herein can represent that the probability of permanent residents and floating population in the initial personnel sub-library appearing in the corresponding administrative region (or the area where the camera is located) is higher than that of other personnel. Figure 6 is an example of the initial appearance probability of personnel in the embodiment of the present application. As Figure 6 shown, the initial appearance probability data of personnel is cycled by time slices. The initial probability data of personnel within each time slice includes the personnel ID, the corresponding time slice, the camera ID, and the probability of the personnel appearing in the camera. For example, within the time slice (0:00, T), the appearance probability of the personnel corresponding to the personnel ID 1 appearing in the camera 1 is 5%, and the appearance probability of the personnel corresponding to the personnel ID 3 appearing in the camera 1 is 10%.

[0102] It should be understood that unless the real-time face capture matching or real-time checkpoint detection of a certain person is triggered subsequently, the initial appearance probability does not change with time.

[0103] In 420, in combination with the historical spatio-temporal trajectory data of personnel, the personnel migration probability distribution matrix within each time slice is determined. Optionally, the historical spatio-temporal trajectory data can be the historical spatio-temporal trajectory data within a period of time, for example, the historical spatio-temporal trajectory data within 1 month, the historical spatio-temporal trajectory data within 1 week, the historical spatio-temporal trajectory data within 3 months, the historical spatio-temporal estimation data within 1 year, etc.

[0104] Specifically, as Figure 7 shown, the input information of the system includes the historical spatio-temporal trajectory data of personnel. The historical spatio-temporal trajectory data of personnel can be the historical spatio-temporal trajectory data of "one person, one file". Among them, "one person, one file" is to establish a personnel file for each person, and the personnel file can include the historical spatio-temporal trajectory data of the corresponding person and other personal information. For example, obtain the activity trajectory data within 1 month in the personnel file of each person.

[0105] In 4201, information such as the personnel ID, the capture time, and the capture camera ID is extracted.

[0106] In 4202, the cameras are associated according to the personnel ID to generate the activity trajectory of each person every day.

[0107] For example, describe the list of cameras passed by the personnel: (camera 1, camera 2,..., camera k).

[0108] For another example, use the method of (personnel ID, time, camera ID) to describe the list of cameras passed by the personnel at each time:

[0109] {PersonID_1, Date_1, weekday_1, (Time_i, Camera_k), (Time_j, Camera_m), …};

[0110] {PersonID_1, Date_2, weekday_2, (Time_n, Camera_l), (Time_s, Camera_r), …};

[0111] …

[0112] {PersonID_x, Date_y, weekday_y, (Time_t, Camera_c), (Time_e, Camera_f), …}.

[0113] In 4203, extract the "one-step" trajectory points of two adjacent cameras in each activity trajectory. The reason for calling it "one-step" is to describe that a person enters the range of subsequent cameras step by step from the range of one camera. Every time a person enters the range of a new camera, it is equivalent to the person "taking one step". In the activity trajectory of the person, these two cameras are in an adjacent relationship.

[0114] For example, the "one-step" trajectory points of the trajectory (Camera 1, Camera 2, Camera 3) are (Camera 1, Camera 2) and (Camera 2, Camera 3). That is to say, a person starts from the shooting range of Camera 1, the next one entered is the shooting range of Camera 2, and then the shooting range of Camera 3 appears.

[0115] In 4204, calculate the personnel migration probability of each "one-step" trajectory point.

[0116] Optionally, based on the activity trajectories of all personnel included in the permanent population database and the floating population database of a certain administrative region, the personnel migration probability of each trajectory point can be statistically obtained.

[0117] In 4205, generate a personnel migration probability distribution matrix within a day with time slices as units.

[0118] Figure 8 Shows an example of the personnel migration probability distribution matrix of an embodiment of the present application. Specifically, Figure 8 Shows the personnel migration probability distribution matrix under each time slice. Taking the time slice (1440 - T, 1440) as an example, the probability that a person migrates from Camera 3 to Camera 1 is 1.2%, and the probability that a person migrates from Camera 3 to Camera N is 3%.

[0119] Optionally, the "one-step" personnel migration probability distribution matrix of the embodiment of the present application is static.

[0120] In 430, based on the face data and checkpoint data captured by the front-end camera, as well as the initial appearance probability of the person generated in 410 and the person migration probability distribution matrix generated in 420, calculate the probability of the person migrating to each camera, that is, update the person appearance probability. Among them, the checkpoint data includes hotel data, travel data, etc.

[0121] Specifically, as Figure 9 shown, the input information of the system includes:

[0122] (1) The initial appearance probabilities of the permanent residents and floating population generated in 410;

[0123] (2) The person migration probability distribution matrix generated in 420;

[0124] (3) The face data captured in real time (for example, person ID, capture time, capture camera ID, matching confidence, etc.), and / or the real-time checkpoint data (for example, person ID, appearance time, appearance location, etc.).

[0125] In 4301, within the time slice [mT, mT + T], the person capture records enter the buffer queue, and the person capture records are traversed and processed in the order of arrival.

[0126] In 4302, determine whether the confidence of the person is higher than the threshold. If it is detected that the confidence of a certain person is higher than or equal to the threshold, execute 4303; if it is detected that the confidence of a certain person is lower than the threshold, then jump back to 4301 and continue traversing.

[0127] Among them, the confidence can be the similarity obtained by performing a 1:N comparison of the person in the population sub-library corresponding to the current camera. That is to say, the probability of the person in the population sub-library corresponding to the current camera.

[0128] In 4303, obtain the probability of the person migrating from the current camera to the surrounding cameras from the person migration probability distribution matrix of the person within the time slice [mT, mT + T].

[0129] In 4304, determine whether the probability of the person migrating from the area where the current camera is located to the surrounding cameras is greater than 0. If this probability is greater than 0, execute 4305; if this probability is equal to 0, then jump back to 4304 and continue traversing.

[0130] In 4305, update the person appearance probability of the surrounding cameras.

[0131] In 4307, load the new list of the person time series sub-libraries in the camera area, that is, update the person set corresponding to each camera.

[0132] Figure 10Shows an example of the appearance probability of updated personnel in an embodiment of the present application. As Figure 10 shown, the personnel IDs in the personnel appearance probability list include the real-name archived activity personnel, and the personnel who moved from the previous hop in the previous time slice to the current time slice (for example, it can be determined according to the personnel migration probability distribution matrix).

[0133] When a certain person appears at the previous hop of the previous time slice of the current time slice, that is, when the person appears at the previous camera corresponding to the camera list shown in Figure 10 , update the probability that the person appears at the camera in the camera list shown in Figure 10 . For example, since personnel IDs 5 - 8 appear at the previous-hop camera of camera 1, the probabilities that personnel IDs 5 - 8 appear at camera 1 are updated to 30%, 25%, 28%, and 26%. Also, for example, personnel IDs 2 and 4 appear at the previous-hop camera of camera 1, and the probabilities that personnel IDs 2 and 4 appear at camera 1 are updated to 15% and 6%. The updated appearance probability values can be obtained based on the confidence level obtained by comparing the personnel IDs 5 - 8 in the corresponding population sub-library of the previous-hop camera and the personnel migration probability of the personnel moving from the previous-hop camera to camera 1.

[0134] When a certain person is confirmed to appear at a certain camera within the time slice (mT - T, mT) (i.e., the previous time slice of the current time slice) and is unlikely to appear at another camera within a time period of duration T, that is, is unlikely to appear at another camera within the time slice (mT, mT + 1), set the probability that the person appears at the other camera to 0. For example, Figure 10 Personnel IDs 1 and 3 in are confirmed to appear at camera 3 and are unlikely to appear at camera 1 within time T, so the probabilities that personnel IDs 1 and 3 appear at camera 1 are set to 0.

[0135] Among them, personnel IDs 1 and 2 are permanent residents among the real-name archived personnel, so personnel IDs 1 and 2 have an initial appearance probability of 10%. Personnel IDs 3 and 4 are floating population among the real-name archived personnel, so personnel IDs 3 and 4 have an initial appearance probability of 15%.

[0136] In Figure 10 , the list of personnel IDs (i.e., the set of personnel) at camera 1 within the time slice (mT, mT + 1) includes personnel IDs 2, 4, 5, 6, 7, 8.

[0137] Understandably, the update of the personnel appearance probability in the embodiments of the present application is triggered by the face data or checkpoint data captured in real time.

[0138] It should also be understood that the corresponding personnel ID list of each camera should include the personnel IDs and face feature information that may be captured by the camera within each time slice. At the same time, it is necessary to avoid including all the personnel information in the entire population database in the ID list, resulting in a degradation to a full-population database traversal comparison.

[0139] In 440, according to the updated personnel appearance probability in 430, a corresponding real-time population accurate sub-database divided by time slices is formed according to the administrative region corresponding to the geographical location of the camera.

[0140] Furthermore, a real-time population accurate sub-database can be formed according to the updated personnel appearance probability in 430, the administrative region corresponding to the geographical location of the front-end camera, and the retrieval logic of personnel filing. Specifically, the personnel IDs that may appear in the current time slice of each camera are added to the corresponding sub-database. Optionally, based on the personnel IDs that may appear in the current time slice of each camera, a face database of the current time slice sub-database can be formed, and feature values are extracted from the face pictures in the sub-database to form an accurate sub-database feature library.

[0141] Figure 11 It is a schematic diagram of the secondary population database sub-database in the embodiments of the present application. It is divided according to the geographical location of the camera in combination with the administrative region. For example, the secondary sub-database is divided by district / county and sub-district office; the primary sub-database is divided by city; the total database volume includes the total permanent population and floating population, generally by province or city.

[0142] Optionally, each level of sub-database can be further divided into one or multiple levels of accurate sub-databases according to the accuracy. Taking the two-level accurate sub-database as an example, the first accurate sub-database includes personnel with a non-zero personnel appearance probability, and the second accurate sub-database includes personnel who have never appeared in the area (that is, the personnel appearance probability is 0) and personnel with a non-zero appearance probability. In this way, when performing personnel comparison, the comparison can be preferentially performed in the first accurate sub-database, further improving the hit rate of the one-time personnel 1:N comparison and reducing the secondary comparison.

[0143] It should be understood that the update of the sub-database list within each time slice needs to be completed within a short time. For example, the time consumption is less than 0.1T.

[0144] Optionally, when face data or checkpoint data enters the cache within the current time slice, after 440, a 1:N comparison of personnel can also be performed based on the population sub-database obtained in 440.

[0145] Figure 12It is a schematic flowchart for personnel comparison based on the population sub-library of the embodiments of the present application. After the front-end camera captures face data, according to the geographical location of the capturing camera, the corresponding population sub-library is associated, and 1:N comparison is performed in this population sub-library to obtain the person with the highest similarity TOP1 to the captured face, and the real-name tag information of this person is obtained, that is, the real-name tag information of the captured person is obtained. In this process, the population sub-library is the population precise sub-library obtained in 440.

[0146] Furthermore, Figure 13 It is a schematic flowchart for a retrieval comparison of the embodiments of the present application. At the current time slice, after the front-end camera captures face data, it will first perform a comparison according to the secondary sub-library corresponding to the area where the camera is located. If the comparison fails, it will further perform a comparison in the primary sub-library corresponding to the area where the camera is located. If it still fails, it will further perform a comparison in the full-scale library. Figure 13 The retrieval logic shown adopts hierarchical comparison of precise sub-libraries, which can improve the comparison accuracy and efficiency and increase the hit rate of personnel in the sub-library.

[0147] Figure 14 It is an implementation form of the system of the embodiments of the present application. Figure 14 Taking the automatic archiving of real-time face capture data of a video surveillance intelligent analysis system as an example, as Figure 14 shown, based on the real-name archived active population library (which can correspond to the historical space-time migration data described above) and personnel flow record data, combined with the real-name permanent population library and floating population library in the administrative region, a real-time regional personnel dynamic precise sub-library is generated; after being imported into the video surveillance intelligent analysis system, through portrait algorithm processing, a face feature value library of the population sub-library is formed; the face capture data (for example, face thumbnail and corresponding scene large picture) of the front-end intelligent camera is uploaded to the video surveillance intelligent analysis system, and the video surveillance intelligent analysis system calls the face algorithm for feature extraction, and maps to the real-time sub-library of the corresponding area in combination with the specific position of the front-end intelligent camera, performs face 1:N comparison, and pushes the comparison result to the personnel file library for archiving.

[0148] The technical solution described above is to dynamically update the personnel ID list corresponding to each camera according to the real-time face data from the camera, and further determine the population sub-library corresponding to each area according to the area where the camera is located, which can ensure that the most active actual population in the area is always maintained in the population sub-library. Compared with the full-scale static library, the invalid personnel data is greatly reduced. When performing the face 1:N comparison real-name tagging process, the hit rate of a single comparison can be improved, and the secondary comparison can be greatly reduced, achieving the effects of less consumption of comparison resources and short time consumption.

[0149] As described above in combination with Figures 1 to 14, which describes in detail the method for dividing the population database provided by the embodiments of the present application. Next, in combination with Figure 15 and Figure 16 , the device embodiments of the present application will be described in detail. It should be understood that the devices in the embodiments of the present application can execute various methods of the foregoing embodiments of the present application, that is, the specific working processes of the following various products can refer to the corresponding processes in the foregoing method embodiments.

[0150] Figure 15 is a schematic structural diagram of the device for dividing the population database according to the embodiments of the present application. Figure 15 The shown device 1500 is only an example, and the devices in the embodiments of the present application may further include other modules or units. As Figure 15 shown, the device 1500 includes a processing unit 1520 and a storage unit 1530.

[0151] The processing unit 1520 is configured to determine N first cameras in the first area, where N is a positive integer.

[0152] The processing unit 1520 is further configured to determine N personnel sets according to the N first cameras, where the probability that the personnel in the i-th personnel set among the N personnel sets appear in the i-th camera is greater than zero, and the i-th personnel set among the N personnel sets corresponds to the i-th camera, and the value of i is each value in [1, N].

[0153] The processing unit 1520 is further configured to determine the first population sub-library of the first area according to the N personnel sets.

[0154] The storage unit 1530 is configured to store the first population sub-library, and the first population sub-library is used to determine the personnel identity information corresponding to the first face data collected by the N first cameras.

[0155] For example, the personnel whose appearance probability in the i-th camera is greater than a certain threshold are included in the personnel set corresponding to the i-th camera. This threshold can be zero or a specific value thereof.

[0156] Optionally, in the embodiments of the present application, different time periods may correspond to different population sub-libraries. For example, a day can be divided into 1440 / T time periods, where T is the duration of each time period, and the 1440 / T time periods can correspond to 1440 / T population sub-libraries.

[0157] In the above technical solution, according to the probability of a person appearing in the first camera, a set of people who may appear in the first camera is determined, and a sub-library of the population database corresponding to this area is determined based on the set of people in each first camera. Since the probability of a person appearing can be continuously updated, it is possible to always keep the most up-to-date actual active population in the sub-library of the population database. Compared with the static sub-library, the amount of invalid personnel data can be greatly reduced. When performing the face 1:N comparison and real-name tagging process, the hit rate of the first comparison can be improved, and the secondary comparison can be significantly reduced, achieving the effects of less consumption of comparison resources and short time consumption.

[0158] In addition, the first sub-library of the population only contains people with a probability of appearance greater than zero. That is to say, the first sub-library of the population only includes people who may appear in the first camera. In this way, the amount of data in the first sub-library of the population can be reduced, and the consumption of comparison resources can be reduced.

[0159] Optionally, the device further includes: an acquisition unit 1510, configured to acquire second face data captured by M second cameras before determining the first sub-library of the population in the first area, where the second cameras are the cameras where the people in the i-th set of people were located before the regional migration. The processing unit 1520 is further configured to determine the probability of the people in the i-th set of people appearing in the i-th camera according to the second face data and the personnel migration probability, where the personnel migration probability is the probability of a person migrating from the second camera to the i-th first camera.

[0160] In the above technical solution, according to the real-time face data from the camera, the probability of a person appearing is updated. That is to say, the list of personnel IDs corresponding to each camera can be dynamically updated. In this way, according to the area where the camera is located, the sub-library of the population database corresponding to each area is determined, and the most up-to-date actual active population in the area can be always kept in the sub-library of the population database. Compared with the full-scale static library, the amount of invalid personnel data is greatly reduced. When performing the face 1:N comparison and real-name tagging process, the hit rate of the first comparison can be improved, and the secondary comparison can be significantly reduced, achieving the effects of less consumption of comparison resources and short time consumption.

[0161] Optionally, the processing unit 1520 is specifically configured to: compare the similarity between the second face data and the face data in the second sub-library of the population to obtain the confidence level of the second face data, where the confidence level is used to indicate the confidence probability that the second face data and the face data in the second sub-library of the population belong to the same person, and the second sub-library of the population is the sub-library of the population corresponding to the second camera; and obtain the probability of the people in the i-th set of people appearing in the i-th camera according to the confidence level and the personnel migration probability.

[0162] Optionally, the confidence corresponding to a certain face data may be multiplied by the personnel migration probability corresponding to the person to obtain the probability of the person appearing.

[0163] Optionally, the processing unit 1520 is further configured to: determine the probability that a person migrates from the second camera to the i-th first camera according to the historical spatio-temporal trajectory data of each person in the permanent population database and / or the floating population database in the first area within a preset time period.

[0164] Optionally, the processing unit 1520 is further configured to: determine the initial probability that the persons in the i-th person set appear in the i-th first camera according to the permanent population database and / or the floating population database in the first area.

[0165] The obtaining unit 1510 may be implemented by a transceiver. The processing unit 1520 may be implemented by a processor. The storage unit 1530 may be implemented by a memory. For the specific functions and beneficial effects of the obtaining unit 1510, the processing unit 1520, and the storage unit 1530, reference may be made to the relevant descriptions in the method embodiments, which will not be elaborated here.

[0166] Figure 16 It is a schematic structural diagram of a sub-population database device provided in another embodiment of the present application. Figure 16 The shown device 1600 (the device 1600 may specifically be a computer device) includes a memory 1601, a processor 1602, a communication interface 1603, and a bus 1604. Among them, the memory 1601, the processor 1602, and the communication interface 1603 are communicatively connected to each other through the bus 1604.

[0167] The processor 1602 is configured to determine N first cameras in the first area, where N is a positive integer; determine N person sets according to the N first cameras, where the probability that the persons in the i-th person set in the N person sets appear in the i-th camera is greater than zero, and the i-th person set in the N person sets corresponds to the i-th camera, and the value of i is each value in [1, N]; determine the first population sub-database of the first area according to the N person sets.

[0168] The memory 1601 is configured to store the first population sub-database, and the first population sub-database is used to determine the personnel identity information corresponding to the first face data collected by the N first cameras.

[0169] Among them, the memory 1601 may be a read only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1601 may store a program. When the program stored in the memory 1601 is executed by the processor 1602, the processor 1602 is configured to execute each step of the method for the population database sub-library in the embodiments of the present application. For example, it may execute Figure 3 Each step of the method embodiment.

[0170] The processor 1602 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method may be completed by the integrated logic circuit in the hardware of the processor or by instructions in software form. The processors described in the embodiments of the present application may be 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. The various methods, steps, and logic block diagrams disclosed in the embodiments of the present application may be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present application may be directly embodied as being executed by a hardware decoding processor, or completed by a combination of hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory (RAM), a flash memory, a read-only memory (ROM), a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory. The processor reads the instructions in the memory and combines its hardware to complete the steps of the above method.

[0171] The steps of the method disclosed in combination with the embodiments of the present application may be directly embodied as being executed by a hardware decoding processor, or completed by a combination of hardware and software modules in the decoding processor. The software module may be located in a random memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory 1601. The processor 1602 reads the information in the memory 1601 and combines its hardware to complete the functions required to be executed by the units included in the population database sub-library device, or execute each method of the method embodiments of the present application.

[0172] The communication interface 1603 uses a transceiver device such as, but not limited to, a transceiver to implement communication between the device 1600 and other devices or communication networks. For example, the face data dynamically captured by the front-end camera can be obtained through the communication interface 1603.

[0173] The bus 1604 may include a path for transmitting information between various components of the device 1600 (e.g., the memory 1601, the processor 1602, the communication interface 1603).

[0174] It should be noted that Figure 16 the processor in the device 1600 may correspond to Figure 15 the processing unit 1520 in the device 1500, the communication interface 1603 may correspond to the acquisition unit 1510, and the memory 1601 may correspond to the storage unit 1530.

[0175] It should be understood that the population database sub-library device shown in the embodiments of the present application may be a server. For example, it may be a server in the cloud, or it may also be a chip configured in a server in the cloud. In addition, the population database sub-library device may be an electronic device, or it may also be a chip configured in an electronic device.

[0176] It should be noted that although the above device 1600 only shows a memory, a processor, and a communication interface, in the specific implementation process, those skilled in the art should understand that the device 1600 may also include other devices necessary for normal operation. At the same time, according to specific needs, those skilled in the art should understand that the device 1600 may also include hardware devices for implementing other additional functions. In addition, those skilled in the art should understand that the device 1600 may also only include the devices necessary for implementing the embodiments of the present application, and do not necessarily include Figure 16 all the devices shown in

[0177] The specific working process and beneficial effects of the device 1600 can be seen in the relevant descriptions in the above method embodiments, and will not be repeated here.

[0178] It should be understood that various aspects or features of the present application can be implemented as a method, an apparatus, or an article of manufacture using standard programming and / or engineering techniques. The term "article of manufacture" as used in the present application encompasses a computer program accessible from any computer-readable device, carrier, or medium. For example, the computer-readable medium may include, but is not limited to: magnetic storage devices (such as hard disks, floppy disks, or magnetic tapes, etc.), optical discs (such as compact discs (CDs), digital versatile discs (DVDs), etc.), smart cards, and flash memory devices (such as erasable programmable read-only memories (EPROMs), cards, sticks, or key drives, etc.). Additionally, the various storage media described herein may represent one or more devices and / or other machine-readable media for storing information. The term "machine-readable medium" may include, but is not limited to, wireless channels and various other media capable of storing, containing, and / or carrying instructions and / or data.

[0179] It should also be understood that in the following embodiments of the present application, "at least one" and "one or more" mean one, two, or more than two. The term " / or" in the present application is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Additionally, the character " / " in this article generally represents an "or" relationship between the associated objects before and after.

[0180] In various embodiments of the present application, the magnitude of the serial numbers of the respective processes does not imply the order of execution. The order of execution of the respective processes should be determined by their functions and internal logics, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0181] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that the computer can access, or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a digital video disc (DVD)), or a semiconductor medium (such as a solid state disk (SSD)), etc.

[0182] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0183] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0184] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0185] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0186] In addition, in each embodiment of the present application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0187] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0188] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present application can easily think of changes or substitutions, which should all be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for partitioning a population database sub - database, characterized in that, it includes: Determine N first cameras in a first area, where N is a positive integer; According to the N first cameras, determine N personnel sets. Among them, the probability that the personnel in the i - th personnel set among the N personnel sets appear in the i - th first camera is greater than zero, and the i - th personnel set among the N personnel sets corresponds to the i - th first camera, and the value of i is each value in [1, N]; According to the N personnel sets, determine the first population sub - database of the first area; Store the first population sub - database, which is used to determine the personnel identity information corresponding to the first face data collected by the N first cameras; Wherein, before determining the first population sub - database of the first area, the method further includes: Obtain second face data captured by M second cameras, where the second cameras are the cameras where the personnel in the i - th personnel set were located before regional migration; According to the second face data and the personnel migration probability, determine the probability that the personnel in the i - th personnel set appear in the i - th first camera, where the personnel migration probability is the probability that a person migrates from the second camera to the i - th first camera.

2. The method according to claim 1, characterized in that, Determining the probability that the personnel in the i - th personnel set appear in the i - th first camera according to the second face data and the personnel migration probability includes: Compare the similarity between the second face data and the face data in the second population sub - database to obtain the confidence level of the second face data. The confidence level is used to indicate the confidence probability that the second face data and the face data in the second population sub - database belong to the same person. The second population sub - database is the population sub - database corresponding to the second camera; According to the confidence level and the personnel migration probability, obtain the probability that the personnel in the i - th personnel set appear in the i - th first camera.

3. The method according to claim 1 or 2, characterized in that, The method further includes: According to the historical spatio - temporal trajectory data of each person in the permanent population database and / or floating population database of the first area within a preset time period, determine the probability that a person migrates from the second camera to the i - th first camera.

4. The method according to claim 1 or 2, characterized in that, The method further includes: According to the permanent population database and / or floating population database of the first area, determine the initial probability that the personnel in the i - th personnel set appear in the i - th first camera.

5. A device for partitioning a population database sub - database, characterized in that, it includes: A processing unit, configured to determine N first cameras in a first area, where N is a positive integer; The processing unit is further configured to determine N sets of personnel according to the N first cameras, where the probability that the personnel in the i-th set of the N sets of personnel appear in the i-th first camera is greater than zero, and the i-th set of the N sets of personnel corresponds to the i-th first camera, and the value of i is each value in [1, N]; The processing unit is configured to determine a first population sub-library of the first area according to the N sets of personnel; The storage unit is configured to store the first population sub-library, and the first population sub-library is used to determine the personnel identity information corresponding to the first face data collected by the N first cameras; Wherein, the device further includes: An acquisition unit, configured to acquire second face data captured by M second cameras before determining the first population sub-library of the first area, where the second cameras are the cameras where the personnel in the i-th set of personnel are located before area migration; The processing unit is further configured to determine the probability that the personnel in the i-th set of personnel appear in the i-th first camera according to the second face data and the personnel migration probability, where the personnel migration probability is the probability that the personnel migrate from the second camera to the i-th first camera.

6. The device according to claim 5, characterized in that, The processing unit is specifically configured to: Compare the similarity between the second face data and the face data in the second population sub-library to obtain the confidence level of the second face data, where the confidence level is used to indicate the confidence probability that the second face data and the face data in the second population sub-library belong to the same person, and the second population sub-library is the population sub-library corresponding to the second camera; Obtain the probability that the personnel in the i-th set of personnel appear in the i-th first camera according to the confidence level and the personnel migration probability.

7. The device according to claim 5 or 6, characterized in that, The processing unit is further configured to: Determine the probability that the personnel migrate from the second camera to the i-th first camera according to the historical spatio-temporal trajectory data of each person in the permanent population library and / or floating population library of the first area within a preset time period.

8. The device according to claim 5 or 6, characterized in that, The processing unit is further configured to: Determine the initial probability that the personnel in the i-th set of personnel appear in the i-th first camera according to the permanent population library and / or floating population library of the first area.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program or instruction, and when the computer program or instruction is executed by the population library sub-library device, the method described in any one of claims 1 to 4 is implemented.

Images