Target tracking method, electronic device and storage medium

Abstract

The present invention discloses a target tracking method, an electronic device and a storage medium, wherein the method comprises: obtaining at least one target in image data, and obtaining file information corresponding to each target, wherein the file information comprises a plurality of related attributes; obtaining a weight value corresponding to each related attribute in the file information, and then using the weight value to determine the danger level of each target; using the danger level to determine a warning target from at least one target, and performing target tracking on the warning target; that is, the present application obtains the file information of the target so that all the targets have corresponding file information, and then uses the file information to determine the danger level of the target, so that the danger level of the target is clear, and then uses the danger level to clarify the warning target and perform target tracking, which can effectively reduce the possibility of missed judgment and misjudgment, and improve the accuracy of identification and tracking.

Description

Technical Field

[0001] The present application relates to the technical field of video surveillance, and in particular to a target tracking method, electronic equipment and storage medium. Background Art

[0002] With the explosive growth of computer applications, video surveillance technology has entered a stage of rapid development. For example, pet management and danger prevention are inseparable from video surveillance technology.

[0003] At present, the phenomenon of keeping pets is increasing, and situations such as pets not being kept on a leash, pets being abandoned in the wild, and pets escaping also occur from time to time. Among them, some pets are fierce pets, have certain aggressiveness, and pose safety hazards. The identification of the dangerousness of pets focuses on identifying unregistered pets, traffic pets, and pets that are not on a leash. The possibility of missed judgments and misjudgments is relatively large, affecting the accuracy of pet identification and tracking. Summary of the invention

[0004] The main technical problem solved by the present application is to provide a target tracking method, electronic device and storage medium. By obtaining the file information corresponding to each target in the image data and obtaining the weight value corresponding to each relevant attribute in the file information, the danger level of each target is determined, and then the danger level is used to clearly identify the warning target and track the target, which can effectively reduce the possibility of missed judgment and misjudgment and improve the accuracy of recognition and tracking.

[0005] To solve the above technical problems, a technical solution adopted in the present application is: to provide a target tracking method, comprising: obtaining at least one target in the image data, and obtaining the file information corresponding to each of the targets, wherein the file information includes multiple related attributes; obtaining the weight value corresponding to each related attribute in the file information, and then using the weight value to determine the danger level of each of the targets; using the danger level to determine a warning target from at least one of the targets, and performing target tracking on the warning target.

[0006] In some embodiments, it also includes: obtaining current archive information and historical archive information corresponding to the target; using the current archive information and the historical archive information to determine target attributes with high similarity; using the target attributes to adjust the target weight value corresponding to the target attributes, and then using the adjusted target weight value to optimize the target tracking.

[0007] In some embodiments, the use of the target attribute to adjust the target weight value corresponding to the target attribute and use the adjusted target weight value to optimize the target tracking includes: in response to the archive level corresponding to the historical archive information being greater than the archive level corresponding to the current archive information, reducing the target weight value; and when the target attribute is a primary attribute, in response to the target weight value being less than or equal to a preset weight value, deleting the target attribute in the current archive information, and then redetermining the danger level of the target to perform the corresponding target tracking of the warning target, wherein the primary attribute is the attribute required when establishing the archive information; or in response to the archive level corresponding to the historical archive information being less than the archive level corresponding to the current archive information, increasing the target weight value; and when the target attribute is a secondary attribute, in response to the target weight value being greater than or equal to a preset weight value, adding the target attribute to the current archive information, and then redetermining the danger level of the target to perform the corresponding target tracking of the warning target, wherein the secondary attribute is an attribute that was not adopted when establishing the archive information but needs to be adopted after the archive information is established.

[0008] In some embodiments, acquiring at least one target in the image data and acquiring archival information corresponding to each of the targets includes: identifying and acquiring at least one of the targets in the image data; acquiring basic attributes of the target, and acquiring matching archival information from an archival information database using the basic attributes.

[0009] In some embodiments, it also includes: in response to a failure in matching the basic attributes of the target with the archival information in the archival information database, obtaining the behavior information and characteristic attributes of the target during the activity; using the behavior information and the characteristic attributes, determining the activity trajectory information of each of the targets; using the basic attributes and the activity trajectory information, establishing corresponding archival information for each of the targets, wherein the archival information determines the archival level according to the number of dangerous attributes in the archival information.

[0010] In some embodiments, the obtaining of the weight value corresponding to each relevant attribute in the archival information, and then using the weight value to determine the danger level of each of the targets, includes: obtaining a statistical value of the number of attribute occurrences of each relevant attribute in the archival information within a preset period; using the statistical value of the number of attribute occurrences to determine the weight value of each of the relevant attributes; using the weight values ​​to construct a danger score matrix, and then using the danger score matrix to determine the danger level of the target corresponding to each of the archival information.

[0011] In some embodiments, it also includes: obtaining the archival level processing items of the archival information, and determining the loss function using the archival level processing items, and then adjusting the weight value using the loss function; in response to the archival level processing items of the archival information reaching a preset processing value, constructing a hazard score matrix using the adjusted weight value; obtaining the fine granularity corresponding to the hazard score matrix, and then determining the classification result of the archival information using the fine granularity; using the hazard score matrix and the classification result, determining the hazard level of the target corresponding to each archival information.

[0012] In some embodiments, the use of the danger level to determine a warning target from at least one of the targets and performing target tracking on the warning target includes: taking the target corresponding to the file information whose danger level is greater than a preset danger level as a warning target; obtaining a target tracking device for the warning target in the current area, and then using the target tracking device to perform target tracking on the warning target.

[0013] In some embodiments, it also includes: obtaining historical trajectory information and current trajectory information of the warning target; using the historical trajectory information and the current trajectory information to determine a target trajectory segment with high similarity; using the target tracking device to complete the trajectory segment to determine the complete trajectory corresponding to the current trajectory information, and then performing target tracking on the warning target in the complete trajectory.

[0014] In some embodiments, it also includes: in response to the current point not obtaining the warning target, obtaining the stay record of the warning target at the current point in the historical trajectory information; in response to the existence of the stay record and the warning target, obtaining the historical stay duration in the stay record, and determining the whereabouts of the warning target based on the historical stay duration.

[0015] To solve the above technical problems, another technical solution adopted in the present application is: to provide an electronic device, the electronic device comprising a memory and a processor coupled to the memory, the memory storing at least one computer program, and the at least one computer program, when loaded and executed by the processor, is used to implement the target tracking method as described above.

[0016] In order to solve the above technical problems, another technical solution adopted in the present application is: providing a computer-readable storage medium, wherein the computer-readable storage medium has at least one program, and when the at least one program is loaded and executed by the processor, it is used to implement the target tracking method as described above.

[0017] Different from the current technology, the target tracking method provided by the present application includes: obtaining at least one target in the image data, and obtaining the archival information corresponding to each target, wherein the archival information includes multiple related attributes; obtaining the weight value corresponding to each related attribute in the archival information, and then using the weight value to determine the danger level of each target; using the danger level to determine the warning target from at least one target, and performing target tracking on the warning target; that is, in the present application, by obtaining the archival information of the target, the target has the corresponding archival information, and then using the archival information to determine the danger level of the target, so that the danger level of the target is clear, and then using the danger level to clarify the warning target and track the target, it can effectively reduce the possibility of missed judgment and misjudgment, and improve the accuracy of identification and tracking. BRIEF DESCRIPTION OF THE DRAWINGS

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

[0019] Figure 1 It is a flowchart of an embodiment of a target tracking method in this application;

[0020] Figure 2 is a schematic diagram of the structure of the image acquisition device in this application;

[0021] Figure 3 This application Figure 1 A flowchart of an embodiment of the working principle of step S10;

[0022] Figure 4 This application Figure 1 A flowchart of an embodiment of the working principle of step S20;

[0023] Figure 5 This application Figure 1 A flowchart of an embodiment of the working principle of step S30;

[0024] Figure 6 It is a flowchart of an embodiment of tracking based on an activity characteristic model in the present application;

[0025] Figure 7 is a schematic diagram of the structure of an embodiment of a target tracking system in the present application;

[0026] Figure 8 It is a structural schematic diagram of an embodiment of an electronic device in the present application;

[0027] Fig. 9It is a structural diagram of an embodiment of a computer-readable storage medium in the present application. DETAILED DESCRIPTION

[0028] The present invention will be further described in detail below in conjunction with the accompanying drawings and examples. It is particularly noted that the following examples are only used to illustrate the present invention, but are not intended to limit the scope of the present invention. Similarly, the following examples are only partial embodiments of the present invention rather than all embodiments, and all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0029] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0030] Traditional target tracking methods, especially for video tracking of pets, usually focus on unregistered pets, traffic pets, and pets that are not on a leash. Pets that meet the above conditions are not necessarily dangerous, and pets that do not meet the above conditions may also be dangerous. Therefore, the possibility of missed judgments and misjudgments is relatively high, affecting the accuracy of target recognition and target tracking of the corresponding pets.

[0031] Therefore, a target tracking method is provided. By obtaining the archival information of each target and the weight value of each relevant attribute in the archival information, the danger level of each target is determined, and then the warning target can be effectively determined from at least one target and the target can be tracked, which can effectively reduce the possibility of missed judgment and misjudgment and improve the accuracy of identification and tracking.

[0032] See also Figure 1 , Figure 1 is a flowchart of an embodiment of the target tracking method in this application; it should be noted that if there are substantial results, the method of this application is not based on Figure 1 The process sequence shown is limited.

[0033] like Figure 1 As shown, the target tracking method may include the following steps.

[0034] S10, obtaining at least one target in the image data, and obtaining file information corresponding to each target, wherein the file information includes a plurality of related attributes.

[0035] Among them, image data can be real-time image data and historical image data obtained in the monitoring area; the target refers to the pet in the image data, such as a dog; the file information refers to the file information established for the pet; the related attributes refer to multiple related attributes of the target in the file information, such as basic attributes and dangerous attributes, among which the basic attributes can be the basic information of the pet, such as breed attributes, color attributes, spot attributes, etc., and the dangerous attributes can be the activity information of the pet, such as abnormal behavior, food protection characteristics, territory characteristics, etc.

[0036] Specifically, an image acquisition device is used to acquire image data of a monitored area to acquire at least one target in the image data, and corresponding archival information corresponding to each target is acquired accordingly, wherein the archival information corresponding to each target may have multiple related attributes, and the related attributes are used to characterize the basic information and activity information of the target.

[0037] S20, obtaining the weight value corresponding to each relevant attribute in the archive information, and then using the weight value to determine the danger level of each target.

[0038] Among them, the weight value refers to the weight value set for each relevant attribute in the file information based on the impact on the danger of the pet; the danger level refers to the danger assessment of the target.

[0039] Specifically, after obtaining the profile information of each target, a weight value is set for each relevant attribute in the profile information using the impact on the danger to the pet, and then the weight value corresponding to each relevant attribute can be used to determine the danger level of each target.

[0040] S30: Determine a warning target from at least one target using the danger level, and perform target tracking on the warning target.

[0041] Among them, warning targets refer to targets whose danger level is greater than the danger threshold and may cause danger.

[0042] Specifically, after determining the danger level of each target, the danger level is compared with the danger threshold, and the target with a danger level greater than the danger threshold is determined as a warning target, and then target tracking is performed on the warning target.

[0043] In this embodiment, by obtaining the archival information of each target in the image data and obtaining the weight value of each relevant attribute in the archival information, the danger level of each target is determined, and then the warning target is determined, and target tracking is performed on the warning target, which can effectively reduce the possibility of missed judgment and misjudgment and improve the accuracy of recognition and tracking.

[0044] In some embodiments, step S10 acquires at least one target in the image data, and acquires the file information corresponding to each target, wherein the file information includes a plurality of related attributes, and may include the following operations.

[0045] First, at least one object in the image data is identified and acquired.

[0046] The target refers to a pet, such as a dog, or other pets that may cause danger. The following example will be given with the target being a dog; the image data may be historical image data or real-time image data acquired in real time.

[0047] Specifically, the image data of the monitored area is acquired by using an image acquisition device, the targets contained in the image data are identified, and then the specific number of targets in the image data is determined.

[0048] Then, basic attributes of the target are obtained, and matching archive information is obtained from an archive information database using the basic attributes.

[0049] Among them, basic attributes refer to the basic information of the pet, such as: breed attributes, color attributes, spot attributes, size attributes, etc.; the archive information database refers to a database storing the archive information of the target, wherein the archive information can be stored in the archive information database in advance.

[0050] Specifically, after acquiring at least one target in the image data, the basic attributes of the target are acquired, and then the basic attributes of the target are compared with the archival information in the archival information database to determine whether there is matching archival information in the archival information database. If there is matching archival information, the successfully matched archival information is extracted as the archival information of the target.

[0051] It can be understood that a successful match indicates that the target in the image data has already established relevant archival information in advance in the archival information database, that is, the current acquisition of archival information of the target is the second acquisition after the second acquisition, and the first acquisition of archival information is used to establish archival information and store it in the archival information database.

[0052] Further, in response to a failure in matching the basic attributes of the target with the profile information in the profile information database, the behavior information and characteristic attributes of the target during the activity are obtained.

[0053] Among them, matching failure means that the target has not yet established file information in the file information database, that is, the file information is obtained for the first time, and the target's file information needs to be established at the first time; behavioral information refers to the behavior of the target during the activity, such as moving, running, chasing, fighting, baring teeth, dancing claws, etc.; characteristic attributes refer to the characteristic attributes of the target under specific circumstances, such as food protection characteristics, territorial characteristics, etc.

[0054] Specifically, when the basic attributes of the target fail to match the profile information in the profile information database, it means that the profile information of the target has not been established in the profile information database. Therefore, the behavior information and characteristic attributes of the current target during the activity are obtained.

[0055] Then use the behavior information and feature attributes to determine the activity trajectory information of each target.

[0056] Among them, activity trajectory information refers to the trajectory information composed of a series of behavior information and characteristic attributes of the target during the activity process.

[0057] Specifically, after obtaining the behavior information and characteristic attributes of the target, the activity track information of each target is determined according to the time relationship between the behavior information and the characteristic attributes on the timeline.

[0058] Then, the basic information and activity trajectory information are used to establish corresponding file information for each target, wherein the file level is determined according to the number of dangerous attributes in the file information.

[0059] Among them, dangerous attributes refer to attributes that affect the danger of the target, such as the breed attribute of fierce dogs in the basic attributes, chasing, baring teeth, and dancing claws in the behavioral information, and the food protection feature in the feature attributes.

[0060] Specifically, after obtaining the basic information and activity track information of the target, the combination of the basic information and the activity track information is used as the basis for establishing the target's profile information, and then corresponding profile information is established for each target.

[0061] In this embodiment, after obtaining the basic attributes of the target, the basic attributes can be used to match the archival information in the archival information database. If the match is successful, the corresponding archival information is directly extracted as the archival information of the target. If the match fails, an archival information is established for the target and stored in the archival information database, so that each target has corresponding archival information, which effectively reduces the possibility of missed judgment and misjudgment.

[0062] In some embodiments, it also includes utilizing the changes in the archive levels of historical archive information and current archive information to adjust the weight value, thereby re-evaluating the danger level of the target and performing subsequent tracking, as described below.

[0063] Get the current archive information and historical archive information corresponding to the target.

[0064] Among them, the current archive information refers to the archive information obtained for the target at the current time, and the historical archive information refers to the archive information obtained for the target before the current time.

[0065] Specifically, the current archive information of the target at the current time is obtained, and the historical archive information of the target stored in the archive information database is obtained.

[0066] Utilize current archive information and historical archive information to determine target attributes with high similarity.

[0067] Among them, the target attribute with high similarity is the attribute that exists in both the current archive information and the historical archive information and has the highest similarity. It can be understood that the target attribute can be one or a set of multiple attributes.

[0068] Specifically, the current archive information and historical archive information of the target are compared to determine the attribute that causes the change in the archive level as the attribute with the highest similarity.

[0069] The target attributes are used to adjust the target weight values ​​corresponding to the target attributes, and then the adjusted target weight values ​​are used to optimize target tracking.

[0070] Further, in response to the archive level corresponding to the historical archive information being greater than the archive level corresponding to the current archive information, the target weight value is reduced.

[0071] Among them, the archive level corresponding to the historical archive information is greater than the archive level corresponding to the current archive information, that is, the archive information representing the target has been downgraded. Therefore, the target weight value corresponding to the target attribute needs to be reduced.

[0072] Specifically, the archive level corresponding to the historical archive information of the target is reduced to the archive level corresponding to the current archive information, that is, the archive level of the target's archive information is downgraded, which means that the target attribute has little impact on the target's dangerousness, so the target weight value of the target attribute can be reduced.

[0073] When the target attribute is the main attribute, in response to the target weight value being less than or equal to the preset weight value, the target attribute is deleted from the current file information, and then the danger level of the target is re-determined to perform target tracking of the corresponding warning target, wherein the main attribute is the attribute required when establishing the file information.

[0074] Specifically, determine whether the target attribute is the main attribute, and when the target attribute is the main attribute, determine whether the target weight value of the target attribute is less than or equal to the preset weight value. If it is less than or equal to the preset weight value, the target attribute can be removed to reduce unnecessary performance consumption. The preset weight value is used to characterize whether the attribute needs to be retained. If it is less than or equal to the preset weight value, the relevant attribute can be removed.

[0075] Or in response to the archive level corresponding to the historical archive information being lower than the archive level corresponding to the current archive information, the target weight value is increased.

[0076] Among them, the archive level corresponding to the historical archive information is lower than the archive level corresponding to the current archive information, that is, the archive information representing the target has been upgraded, therefore, the target weight value corresponding to the target attribute needs to be increased.

[0077] Specifically, the archive level corresponding to the historical archive information of the target is upgraded to the archive level corresponding to the current archive information, that is, the archive level of the target's archive information is upgraded, indicating that the target attribute has a greater impact on the target's dangerousness, so the target weight value of the target attribute can be increased.

[0078] When the target attribute is a secondary attribute, in response to the target weight value being greater than or equal to the preset weight value, the target attribute is added to the current file information, and then the danger level of the target is re-determined to perform target tracking of the corresponding warning target, wherein the secondary attribute is an attribute that is not used when the file information is established but needs to be used after the file information is established.

[0079] Specifically, determine whether the target attribute is a primary attribute, and if the target attribute is a non-primary attribute, determine whether the target weight value of the target attribute is greater than or equal to the preset weight value. If it is greater than or equal to the preset weight value, add the target attribute to the current file information, and then redetermine the weight value of each attribute in the file information, and then redetermine the danger level of the target to perform target tracking of the warning target.

[0080] In this embodiment, the file level of the target's file information is lowered or raised, and the weight values ​​of the relevant attributes are adjusted accordingly, thereby re-determining the danger level of the target, and corresponding target tracking of the warning target is performed, thereby further improving the accuracy of target recognition and target tracking.

[0081] In some embodiments, the image acquisition device is as follows:

[0082] See also Figure 2 , Figure 2 It is a structural schematic diagram of the image acquisition device in this application.

[0083] like Figure 2As shown, the image acquisition device includes an image capture module, an image analysis module, a dynamic adjustment module, a management module and a storage module. Among them, the image analysis module can include a leash analysis component, an abnormal behavior analysis component, a variety analysis component, and an expandable component; the dynamic adjustment module can include a downgraded archive attribute analysis component, an upgraded archive attribute analysis component, and an upgraded archive feature attribute learning component. Among them, the image capture module is used to acquire image data, and the image analysis module is used to perform leash analysis of the leash analysis component, abnormal behavior analysis of the abnormal behavior analysis component, variety analysis of the variety analysis component, and analysis of the expandable component on the image data; then the analysis results are transmitted to the management module for management, and then transmitted to the storage module for storage. The storage component can also be connected to the outside to transmit the stored data to the outside; then based on the further analysis structure of the analysis results, the dynamic adjustment module is used to perform attribute analysis or learning of the corresponding component, and then adjust the acquisition conditions and feedback to the image analysis module.

[0084] The working principle of step S10 is shown in Figure 3 , as shown below.

[0085] Within the monitoring area, the initial image analysis model is used to identify the target (dog) that appears, and then the corresponding archive information is established. In a certain period, the upgraded or downgraded archive information is evaluated by extracting relevant attributes, and then adaptively increasing or decreasing the collection dimensions of relevant attributes to match the needs. The specific operation steps are as follows:

[0086] Analyze the target (dog) in the image data to identify whether the target is on a leash. If not, use image comparison technology to compare the records in the archive information database. If the archive information of the target does not exist in the archive information database, capture the image of the target in the image data, upload it to the archive information database, and record it as the first target archive information, i.e., dog. file.

[0087] If it is on a leash, the target type is analyzed to see if it is a fierce dog. If it is a fierce dog and there is no corresponding file information in the file information database, it is recorded as the second target file information, that is, dog type file.

[0088] If the dog is on a leash and is not a fierce dog, analyze the target's behavior and if there is any abnormal behavior or dangerous habits For example, if more than y off-leash persons are tracked, rabies behavior is matched, etc., and there is no corresponding profile information in the profile information database, then it is recorded as the third target profile information, i.e., dog file.

[0089] In some embodiments, the collection module can also be expanded to formulate rules according to actual conditions, such as target color, target habits, etc., and when the rules are met and the corresponding file information does not exist in the file information database, it is recorded as the nth target file information, that is, dog file.

[0090] In addition, you can also obtain the basic attributes of potentially dangerous dogs , such as the dog's breed, color, characteristic attributes (spots), etc.

[0091] For dogs that have been archived and marked as dangerous, record the capture points , and judge whether it is a densely populated area (for example, the number of people in the image data exceeds y), if it meets the requirement, it will be recorded ; Capture time , the duration of stay at the current point .

[0092] Record the abnormal behaviors and characteristic attributes of the target's activities at each time point, and record abnormal behaviors , record dangerous habits (such as food protection characteristics, territorial characteristics, and record the leash as this attribute when creating the file) , as a snapshot activity track record record( ), represents an activity track record, Represents a collection of abnormal behaviors, Represents a collection of dangerous habits, Whether the representative appears in crowded areas, It's the capture point. It's capture time. is the stay time; then a corresponding file of each target is formed, and the target's file information is recorded as .

[0093] Set a statistical period. (e.g. y days), if the number of files processed reaches a certain amount , then extract the archive information of all processed archives and record it as .

[0094] Information about all downgraded profiles , for the archive information that has been downgraded to the same archive level and the original archive level before downgrading, find the attribute with the highest similarity (the attribute can be a single attribute or a set containing multiple attributes), adjust the factor coefficient, and calculate using the variance analysis method:

[0095] ;

[0096] Among them, sh represents Each dangerous attribute of the file, , that is, calculate the F statistic for each dangerous attribute at the current level and take the largest value (the larger the value, the greater the impact of the dangerous attribute on the overall score); then Calculated together with the archives of the same level after downgrading:

[0097] ;

[0098] in, Refers to removing from the current level After the file is saved, the attribute with the greatest impact in the remaining file; max( ) means to remove After the file is created, the attribute with the greatest impact in the original level file is left. The purpose of the operation is to find out which attribute caused the level change; then calculate the original level file:

[0099] ;

[0100] Among them, if

[0101] ;

[0102] and

[0103] ;

[0104] Then reduce the influence weight of this attribute If the attribute is the main attribute of the collection, determine , if true, remove the condition from the acquisition logic, Refers to the weight value in the attribute algorithm; set The purpose of is to evaluate whether the attribute as the collection condition still needs to be retained. If it is lower than this value, the attribute can be removed from the collection condition to reduce unnecessary performance consumption of the system.

[0105] For all updated archive information , upgrade to the same level, the original level before the upgrade, respectively obtain the most similar attributes, and adjust the factor coefficient. The same variance analysis method is used to calculate:

[0106] ;

[0107] Among them, Calculated together with the upgraded files of the same level:

[0108] ;

[0109] Recalculate the original level file:

[0110] ;

[0111] like

[0112] ;

[0113] and

[0114] , then increase the influence weight of this attribute , that is, to increase the target weight value corresponding to the target attribute. If the attribute is not the main attribute of the collection, determine , if it is true, then add this condition to the collection logic, that is, add this attribute.

[0115] Furthermore, if a picture feature extraction model is provided, it supports extracting all picture features of the same type of files and supplements the collection conditions. Extract common feature attributes from relevant archive pictures. If the verification accuracy is greater than , then an algorithm module for collecting specific attribute conditions is formed to supplement the collection conditions.

[0116] In some embodiments, step S20 obtains the weight value corresponding to each relevant attribute in the archive information, and then uses the weight value to determine the danger level of each target, which may include the following operations.

[0117] First, obtain the attribute occurrence statistics of each relevant attribute in the archive information within a preset period.

[0118] Among them, the preset period refers to a cycle time, and period = 1period refers to the cycle time of a preset period, which can be 5 days, 7 days, 10 days, etc., and can be set according to actual conditions; the attribute occurrence count statistics refer to the occurrence count of each relevant attribute in the image data within the preset period.

[0119] Specifically, after acquiring the target's archival information, multiple image data collections are performed on the target within a preset period to obtain statistics on the number of occurrences of each relevant attribute of the target within the preset period, and obtain attribute occurrence statistics of each relevant attribute.

[0120] In some embodiments, in order to more accurately determine the danger level of a target, attributes related to potential danger may be extracted as danger attributes, and then attribute occurrence statistics corresponding to the danger attributes may be obtained.

[0121] Next, the attribute occurrence statistics are used to determine the weight value of each relevant attribute.

[0122] Among them, the larger the statistical value of the number of occurrences of the relevant attribute, the more times the relevant attribute appears, and the greater the impact on the dangerousness of the target. It can be understood that in order to more accurately determine the danger level of the target, the relevant attributes of potential danger can be extracted as dangerous attributes, that is, the relevant attributes mentioned here can be dangerous attributes.

[0123] Specifically, after obtaining the attribute occurrence statistics of each relevant attribute, the relevant attributes of potential danger, namely the dangerous attributes, are extracted; and then the weight value of each dangerous attribute is calculated using the attribute occurrence statistics of the dangerous attributes, and the weight value can be calculated using the entropy weight method.

[0124] Then, the weight values ​​are used to construct a risk score matrix, and the risk score matrix is ​​used to determine the risk level of the target corresponding to each file information.

[0125] Among them, the purpose of constructing a hazard score matrix is ​​to comprehensively judge multiple hazard attributes to obtain the hazard score corresponding to each file information, and then determine the hazard level of the target based on the hazard score.

[0126] Specifically, after obtaining the weight values ​​of relevant attributes, the weight values ​​of dangerous attributes are used to construct a danger score matrix, and then the danger score corresponding to each piece of file information is determined. Then, based on the danger score, the danger level of each piece of file information is determined, that is, the danger level of each target is determined.

[0127] Furthermore, a feedback adjustment mechanism may be included, that is, adjusting the corresponding weight value according to the archive level of the archive information, as follows:

[0128] Obtain the archive level processing items of the archive information, and use the archive level processing items to determine the loss function, and then use the loss function to adjust the weight value.

[0129] Among them, archive grade processing matters refer to the adjustment of the archive grade of archive information, which can be an upgrade or downgrade.

[0130] Specifically, after obtaining the archive level processing items of the archive information, the loss function is determined using the number of grade adjustments and the total number of archive level processing, and then the weight value is adjusted using the loss function.

[0131] In response to the archival level processing items for archival information reaching a preset processing value, a risk score matrix is ​​constructed using the adjusted weight values.

[0132] Among them, the preset processing value is a preset threshold value of the archive level processing items. It can be understood that when the archive level processing items reach the preset processing value, the feedback adjustment mechanism is triggered.

[0133] Specifically, when the archive level processing items of the archive information reach the preset processing value, the feedback adjustment mechanism is triggered, and the weight values ​​obtained after adjustment are used to construct a risk score matrix.

[0134] The fine granularity corresponding to the hazard score matrix is ​​obtained, and then the classification result of the archival information is determined using the fine granularity.

[0135] Among them, the classification result refers to the classification based on the existing data, which can be classified using an unsupervised classification algorithm.

[0136] Specifically, after obtaining the risk score matrix, the fine granularity corresponding to the risk score matrix is ​​obtained, and then the clustering algorithm is used to classify the archive information according to the fine granularity corresponding to the risk score matrix to determine the classification result of the archive information.

[0137] It can be understood that the classification result of each archive information includes at least one category and no more than K categories, and K can be 4.

[0138] Using the hazard score matrix and classification results, determine the hazard level of the target corresponding to each file information.

[0139] Specifically, after obtaining the hazard score matrix and classification results, determine the hazard level of the target corresponding to each file information; the hazard level can be divided into low hazard, medium hazard, high hazard and extremely high hazard, and the classification result of each file information can be one or more of low hazard, medium hazard, high hazard and extremely high hazard.

[0140] In this embodiment, the danger level of the archive information is determined by the weight values ​​of the relevant attributes, so that the danger level corresponding to each target can be clearly identified, thereby improving the accuracy of identification.

[0141] The working principle of step S20 can be found in Figure 4 , as shown below.

[0142] By extracting The attributes related to potential danger in the archives, that is, extracting the relevant attributes of potential danger in the archive information, assigning weights to the attributes and adaptively adjusting them in combination with the negative feedback mechanism, calculate and obtain more reasonable potential danger level assessment and suggestions, as follows.

[0143] Using prior knowledge from pet science, zoology, sociology, etc., we extract risk-related factors from the basic attributes of dogs, such as breed, size, age, and gender. ,in .

[0144] right In Conduct analysis and design statistical cycles (such as y day), yes Sum the number of abnormal behaviors in ; ( ) represents an activity track record. Represents a collection of abnormal behaviors. Represents the statistics of the number of abnormal news in a period from day y to yesterday.

[0145] right In Analyze and design statistical methods (such as daily statistics, if multiple captures are made in one day, only record one time, and count as one cycle). (such as y days)), that is ,in, ; Represents a collection of dangerous habits, It means that for a dangerous behavior, it can be captured once within t days (because a dangerous behavior may be captured repeatedly in a day, so frequent recording is not very meaningful). , represents the statistics of the number of dangerous habits in a cycle from day y to yesterday.

[0146] right In Conduct analysis and design statistical cycles (e.g. y days), the stay time meets certain conditions (e.g. greater than minutes) to sum up the number of ,in ; Whether it appears in a crowded area, greater than Minutes can be set according to user specific usage, , represents the statistics of the number of times people stayed in densely populated areas within a period from day y to yesterday.

[0147] right In ( ) to analyze and design statistical cycles (such as y days), for the historical trajectory ( ) Combine statistics to obtain high-frequency points and time periods ( ), design a priori rules, such as and , then the probability that the dangerous dog is a stray animal is high, and stray=1; if 17≤ and ,at the same time If there is no such record, the probability that the dangerous dog is raised by humans is relatively high, and stray=0; ( )middle It's the capture point. It's capture time. It is the length of stay; ( ) represents the high-frequency point and the time period in which it appears, where It is a high frequency point. It is the high-frequency time period, that is, which time period in 24 hours is used (0~24 represents 00:00:00 respectively).

[0148] In some embodiments, the extended attributes can also be obtained based on the self-learning ability of the acquisition module, which is recorded as .

[0149] It is understandable that the factors mentioned in the above steps have a positive impact on the risk level, so there is no need to specially process the data. If there are negative indicators after expansion, they need to be positive. The dimensions of each factor are different, so all factors are standardized: the parameters of the entropy weight method have provided calculation rules, which will not be described in detail, and then:

[0150] ;

[0151] Among them, the standardized factors are used to calculate the weight of each factor using the entropy weight method. First calculate the entropy value: , and then calculate the weight, ;

[0152] Then calculate the potential risk score matrix corresponding to the archive information, .

[0153] Then calculate the risk score of each file information

[0154] .

[0155] Because this program collects dogs with potential danger, the classification of their danger levels is relative. Unsupervised classification algorithms are used to classify them according to existing data, which is adaptive. The fine granularity will The archives are classified unsupervised to form c categories, which are no more than 4 categories, namely .

[0156] Each type of file is assigned a danger level identification that matches the user's identification: The automatic solution is as follows: Because this solution collects dogs that are potentially dangerous, When, according to size, giving "medium" and "high" grades; when When , it is assigned a "high" level; when When , assign "low", "medium" and "high" grades; when , it is given an "extremely high" level, that is:

[0157] .

[0158] Support by setting The size of the hazard level can be manually configured, such as:

[0159] .

[0160] Furthermore, a negative feedback adjustment mechanism is provided; the user clicks "downgrade" or "upgrade" after viewing the profile as an evaluation indicator to adjust the attribute weight. (e.g. y days) later, if the number of files processed by the user reaches a certain amount , then the feedback adjustment mechanism is triggered, see the working principle of step s10, and then the objective function is designed. ,pass The result is to use the Adam algorithm to optimize the weight value. This is the gradient correction and is recorded in subsequent calculations.

[0161] In some embodiments, step S30 determines a warning target from at least one target using the danger level and performs target tracking on the warning target, which may include the following steps.

[0162] First, the target corresponding to the archive information with a danger level greater than the preset danger level is used as a warning target.

[0163] Among them, the preset danger level is a dividing line for dividing warning targets, which can be set according to actual conditions. For example, if the medium danger level is set as the preset danger level, the target corresponding to the file information with a danger level greater than the medium danger level will be used as the warning target, that is, the target corresponding to the file information with high danger and extremely high danger will be used as the warning target; or the target of the file information with a danger score greater than the preset danger score will be set as the warning target. The preset danger score can be set according to actual conditions. For example, if the preset danger score is set to 60 points, the target of the file information with a danger score greater than 60 points will be used as the warning target.

[0164] Obtain a target tracking device for the warning target in the current area, and then use the target tracking device to perform target tracking on the warning target.

[0165] The current area refers to the area where the warning target is located, and the target tracking device refers to the image data acquisition device.

[0166] Specifically, after obtaining the warning target, the current area of ​​the warning target during movement is obtained, and the target tracking device in the current area is obtained, and then the target tracking device is used to perform target tracking on the warning target.

[0167] It is understandable that when the warning target is moving, the current area will also move with the warning target, that is, the current area is the area centered on the warning target, that is, the target tracking device can be a target tracking device within a certain range during the movement of the warning target.

[0168] Furthermore, it also includes:

[0169] Get the historical and current trajectory information of the warning target.

[0170] The historical trajectory information refers to the activity trajectory information contained in the archive information stored in the archive information database, and the current trajectory information refers to the activity trajectory information contained in the archive information currently obtained.

[0171] Specifically, the activity track information of the warning target contained in the archive information stored in the archive information database, that is, the historical track information, is obtained, and the current activity track information of the warning target, that is, the current track information, is obtained.

[0172] The historical trajectory information and the current trajectory information are used to determine the target trajectory segments with high similarity.

[0173] The historical trajectory information represents the trajectory that the target has passed through during historical activities, the current trajectory information represents the trajectory that the target has passed through during current activities, and the trajectory segment refers to a segment of the trajectory in the trajectory information, that is, a part of the trajectory information.

[0174] Specifically, after obtaining the historical trajectory information and the current trajectory information, the historical trajectory information and the current trajectory information are compared to find the target trajectory segments with high similarity; it can be understood that the target trajectory segments with high similarity refer to, with respect to streets, passing through the same streets. For example, if the historical trajectory information contains wandering back and forth on the same street, and the current trajectory information contains going straight on the same street, these also belong to the target trajectory segments with high similarity.

[0175] The target tracking device is used to complete the trajectory fragment to determine the complete trajectory corresponding to the current trajectory information, and then target tracking is performed on the warning target in the complete trajectory.

[0176] Among them, there may be cases where the target is missing in the trajectory information, such as being blocked or returning home. Therefore, after obtaining trajectory segments with high similarity, the historical trajectory information and the current trajectory information can be compared to determine the target's habitual trajectory and then complete the trajectory segments.

[0177] Specifically, after obtaining a target trajectory segment with high similarity, the target tracking device corresponding to the target is used to search and obtain the target's current trajectory, and then complete the estimated segment to determine the complete trajectory corresponding to the current trajectory information, and then perform target tracking on the warning target in the complete trajectory.

[0178] Furthermore, it also includes:

[0179] In response to the fact that the warning target is not obtained at the current point, a stay record of the warning target at the current point in the historical track information is obtained.

[0180] Among them, the warning target is not obtained at the current point, that is, the target is blocked or returns home. Therefore, it is necessary to obtain the stay record of the warning target at the current point in the historical trajectory information to distinguish whether the target is blocked or returns home.

[0181] Specifically, when the warning target is not obtained at the current point in the current trajectory information, it is determined whether there is a stay record of the warning target at the current point in the historical trajectory information, and if so, the stay record of the warning target at the current point in the historical trajectory information is obtained.

[0182] In response to the existence of the stay record and the warning target, the historical stay duration in the stay record is obtained, and the whereabouts of the warning target is determined based on the historical stay duration.

[0183] Among them, the historical stay time can represent whether the target is blocked or returns home at the current point.

[0184] Specifically, when the warning target exists at the current point in the historical trajectory information and there is a stay record at the current point, the historical stay duration of the warning target in the stay record is obtained, and the historical stay duration is used for judgment. When the historical stay duration is greater than the stay threshold, it is determined that the warning target has returned home. When the historical stay duration is less than the stay threshold, it is determined that the warning target is blocked.

[0185] The working principle of step S30 can be found in Figure 5 , as follows:

[0186] For the files whose danger levels have been analyzed in the working principle of step S20, the activity habits are analyzed, and the activity habits are predicted, including but not limited to high-frequency appearance time points and locations, periodic long-term stay locations, temporary long-term stay locations, etc., and pushed to relevant personnel for confirmation. Relevant personnel can downgrade or upgrade the danger level according to regional characteristics. After determining the tracking target, an adaptive sliding device topology network is established through historical trajectories and habits, and linkage tracking is carried out. Finally, relevant personnel are notified for collaborative governance. The specific steps are as follows:

[0187] Set push thresholds, such as "high" or "extremely high" risk levels, or files with a score greater than a certain value.

[0188] For high-frequency haunting points and haunting periods ( ) archives, using time series models such as ARIMA to predict the locations of points that appear within a specific time period .

[0189] Proactively alert potential dangerous dogs that are about to appear within a specific time. and archival information , use information transmission devices such as apps to push them to relevant personnel.

[0190] After receiving the information, relevant personnel can confirm that the information content is correct and choose to maintain, downgrade or upgrade the danger level; after confirming the dangerous situation, they can choose whether to track the target.

[0191] After confirming the target is being tracked, the device at the current location and the devices in the surrounding locations will be linked to start the tracking mode.

[0192] Recorded by this dog ( ) for analysis, with residence time The location where the track is cut is the end point. and current point The trajectory of the same trajectory is extracted by using the method of solving the trajectory similarity. If the number of points in the current segment is less than 3, it is removed. After the time point, obtain the trajectory segment with the highest similarity again; if it is still less than 3, take the fixed number of the nearest front and rear devices in the spatial relationship to complete the trajectory segment; and it is necessary to ensure that the current point If it is not at the starting point or the end point of the track segment, then the track segment that is not cut is retrieved and the front / back point is completed. Finally, , that is, the complete trajectory is obtained.

[0193] At the current point As the center point, contrast and ( )middle The probability of the points before and after the point appearing, if the probability is greater than , then replace middle The front and rear points are formed .extract About A fixed number of points before and after are set to form a topological network, and the replacement points are used as backup points.

[0194] Get of The video streams of the front and rear points based on the center point, as well as the video streams of the backup points. For the video stream frames in the topology network, different features are extracted based on the image of the previous frame and the image of the current frame, and the features are compared with the previous and next frames of the devices between networks to reduce the calculation pressure of similarity comparison; after reaching the similarity threshold, the automatic tracking is successful, the direction of travel is analyzed and output, and the next device is driven to actively focus on the target, and the information that the target is at this point is reported to the relevant users; the similarity threshold needs to be manually set based on the overall recognition situation after the device is deployed on site, for example, the similarity can be set to 90%. After the tracking is successful, It becomes the next device that has been successfully tracked and repeats the above operations to achieve automatic tracking of each device one by one.

[0195] If it appears midway The track segment of interest in the current track does not capture the moving object, that is, the target is likely to remain stationary, or has hidden, or escaped from the tracking link; The point video stream is analyzed, and features are obtained based on the target's archive records (based on basic attributes and captured images). The similarity of the target in the video stream of the previous and next frames is analyzed, and the stay record of the point in the historical trajectory is analyzed; if the target exists and the record exists, and there is a long-term stay record in the record, then , the tracking link always focuses on the point, and the reported information stays, and the clustering algorithm is used for all Cluster analysis is used to obtain the value of the center point of the largest class's dwell time, which is used as the predicted dwell time to report to the relevant personnel. After waiting for the historical minimum dwell time, real-time analysis begins. If the target exists but the record does not exist or the threshold is not met, the video stream is still analyzed in real time. If the target does not exist but the record exists and the threshold condition is met, the dwell time is also predicted, and the target is hidden and the location information is reported to the relevant personnel. If neither the target nor the record exists, the target is determined to be lost, and the lost location is recorded and reported to the relevant personnel. See the attached tracking process. Figure 6 The historical minimum stay time refers to the time when the object has been judged as medium, high, or extremely high risk, and the captured trajectory is recorded in real time and stored in the archive. After the recorded trajectory in the archive matches the current similar trajectory, the entire archive is searched for the length of time it stays at the point, and the minimum length of time is taken; the threshold condition here refers to the above-mentioned stay time .

[0196] After successful tracking, relevant personnel will go to the target location for management. After completion, the file will be marked as managed, the file danger level will be adjusted to "low", and it will be deleted regularly.

[0197] In some embodiments, a tracking process diagram based on the activity characteristic model is also included.

[0198] See also Figure 6 , Figure 6 It is a flowchart of an embodiment of tracking based on activity characteristic model in this application.

[0199] like Figure 6 As shown in the figure, the historical trajectory is cut; then similar trajectories are obtained; the trajectory is adjusted according to the predicted potential; a topological network between devices is established according to the trajectory; the video stream between the topological networks is analyzed; it is determined whether the similarity of the moving object is greater than the threshold. If it is greater than the threshold, the target tracking is successful, and the device actively focuses on the target; then the point and information are reported to the relevant personnel; if the similarity of the moving object is not greater than the threshold, the active characteristic model is used for analysis, specifically: based on the similarity analysis of the archive information, the historical stay record analysis, and whether the target exists. If the target exists, it is determined whether the historical stay time is greater than the threshold. If the historical stay time is greater than the threshold, the clustering prediction stay time is used, and then After waiting for a fixed time, the focus point is analyzed in real time, and then the point and information are reported to relevant personnel; if the historical residence time is not greater than the threshold, real-time analysis is performed, and then the point and information are reported to relevant personnel; if the target does not exist, it is determined whether the historical residence time is greater than the threshold. If the historical residence time is greater than the threshold, the residence time is predicted by clustering, and it is determined that the target is hidden. Then the focus point is waited for a fixed time and then real-time analysis is performed, and then the point and information are reported to relevant personnel; if the historical residence time is not greater than the threshold, it is determined that the target is lost, and then the point and information are reported to relevant personnel; it is determined whether the tracking is ended, if so, the process ends, otherwise it returns to the step of cutting the historical track.

[0200] In this embodiment, a warning target is determined from at least one target by the danger level, and then target tracking is performed by the target tracking device of the warning target in the current area. If the target is obscured, the target trajectory fragment with high similarity is determined through the historical trajectory information and the current trajectory information, and then the complete trajectory corresponding to the current trajectory information is completed, and the historical stay time is determined based on the stay record of the warning target at the current point, and then it is distinguished whether the warning target is obscured or returning home, so as to complete the accurate tracking of the warning target and reduce the occurrence of missed judgments and misjudgments.

[0201] In some embodiments, a target tracking system is also provided.

[0202] See also Figure 7 , Figure 7 It is a structural diagram of an embodiment of a target tracking system in the present application.

[0203] like Figure 7As shown, the target tracking system 400 includes: a first acquisition module 410, a second acquisition module 420 and a determination module 430; wherein the first acquisition module 410 is used to acquire at least one target in the image data, and acquire the file information corresponding to each target, wherein the file information includes a plurality of related attributes; the second acquisition module 420 is used to acquire the weight value corresponding to each related attribute in the file information, and then use the weight value to determine the danger level of each target; the determination module 430 uses the danger level to determine the warning target from at least one target, and performs target tracking on the warning target.

[0204] In some embodiments, an electronic device is also provided.

[0205] See also Figure 8 , Figure 8 1 is a schematic diagram of the structure of an electronic device in an embodiment of the present application. The electronic device can execute the steps of the target tracking method in the above method.

[0206] The electronic device 500 includes a memory 520, a processor 510 coupled to the memory, and at least one computer program stored in the memory 520 and executable on the processor 510; when the processor 510 loads and executes the at least one computer program, it is used to implement the steps of the target tracking method in the above method. For related contents, please refer to the detailed description in the above method, which will not be repeated here.

[0207] In some embodiments, a computer-readable storage medium is also included.

[0208] See also Fig. 9 , Fig. 9 It is a structural diagram of an embodiment of a computer-readable storage medium in the present application.

[0209] The computer readable storage medium 600 stores at least one program 610, which is used to implement the steps of the target tracking method in the above method when loaded and executed by the processor. For related contents, please refer to the detailed description of the above method, which will not be repeated here.

[0210] The above scheme obtains the archival information of the target in the image data and the weight value of each relevant attribute in the archival information, and then uses the weight value to determine the danger level of each target, and uses the danger level to determine the warning target, and then performs target tracking. It can effectively use the archival information of the target in the image data to accurately determine the danger level of each target, which can effectively reduce the possibility of missed judgment and misjudgment and improve the accuracy of recognition and tracking.

[0211] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device implementation described above is only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as 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 mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0212] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0213] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0214] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0215] The above description is only an implementation mode of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A target tracking method, characterized in that: include: Acquire at least one target in the image data, and acquire file information corresponding to each of the targets, wherein the file information includes a plurality of related attributes; Obtaining a weight value corresponding to each relevant attribute in the archival information, and then using the weight value to determine the danger level of each of the targets; wherein, it also includes: obtaining current archival information and historical archival information corresponding to the target; using the current archival information and the historical archival information to determine target attributes with high similarity; using the target attributes to adjust the target weight value corresponding to the target attribute; wherein, in response to the archive level corresponding to the historical archival information being greater than the archive level corresponding to the current archival information, reducing the target weight value; or in response to the archive level corresponding to the historical archival information being less than the archive level corresponding to the current archival information, increasing the target weight value; and then using the adjusted target weight value to re-evaluate the danger level of each of the targets; A warning target is determined from at least one of the targets using the danger level, and target tracking is performed on the warning target.

2. The method according to claim 1, characterized in that The reducing the target weight value further includes: When the target attribute is a primary attribute, in response to the target weight value being less than or equal to a preset weight value, the target attribute is deleted from the current archive information, and the danger level of the target is re-determined to perform target tracking of the corresponding warning target, wherein the primary attribute is an attribute required when establishing the archive information; The step of increasing the target weight value further includes: When the target attribute is a secondary attribute, in response to the target weight value being greater than or equal to a preset weight value, the target attribute is added to the current file information, and then the danger level of the target is re-determined to perform target tracking of the corresponding warning target, wherein the secondary attribute is an attribute that is not used when establishing the file information but needs to be used after the file information is established.

3. The method according to claim 1, characterized in that The acquiring of at least one target in the image data and acquiring the archive information corresponding to each of the targets includes: identifying and acquiring at least one of the targets in the image data; The basic attributes of the target are obtained, and matching archive information is obtained from an archive information database using the basic attributes.

4. The method according to claim 3, characterized in that Also includes: In response to a failure in matching the basic attributes of the target with the profile information in the profile information database, obtaining behavior information and characteristic attributes of the target during the activity; Determine the activity track information of each of the targets by using the behavior information and the characteristic attributes; Corresponding file information is established for each of the targets using the basic attributes and the activity track information, wherein the file level of the file information is determined according to the number of dangerous attributes in the file information.

5. The method according to claim 1, characterized in that The obtaining of a weight value corresponding to each relevant attribute in the archive information, and then using the weight value to determine the danger level of each target, includes: Obtaining a statistical value of the number of occurrences of each relevant attribute in the archive information within a preset period; Determine the weight value of each of the related attributes using the attribute occurrence statistics; The weight values ​​are used to construct a risk score matrix, and the risk score matrix is ​​then used to determine the risk level of the target corresponding to each file information.

6. The method according to claim 5, characterized in that Also includes: Obtaining the archive level processing items of the archive information, and using the archive level processing items to determine a loss function, and then using the loss function to adjust the weight value; In response to the archival level processing items of the archival information reaching a preset processing value, constructing a risk score matrix using the adjusted weight values; Obtaining the fine granularity corresponding to the risk score matrix, and then using the fine granularity to determine the classification result of the archive information; The hazard level of the target corresponding to each file information is determined by using the hazard score matrix and the classification result.

7. The method according to claim 1, characterized in that The step of determining a warning target from at least one of the targets by using the danger level, and performing target tracking on the warning target, includes: The target corresponding to the archive information whose danger level is greater than the preset danger level is used as a warning target; A target tracking device of the warning target in the current area is obtained, and then the target tracking device is used to perform target tracking on the warning target.

8. The method according to claim 7, characterized in that Also includes: Obtaining historical trajectory information and current trajectory information of the warning target; Determine a target trajectory segment with high similarity by using the historical trajectory information and the current trajectory information; The target tracking device is used to complete the trajectory segment to determine a complete trajectory corresponding to the current trajectory information, and then target tracking is performed on the warning target in the complete trajectory.

9. The method according to claim 8, characterized in that Also includes: In response to the current point not acquiring the warning target, acquiring a stay record of the warning target at the current point in the historical trajectory information; In response to the stay record and the existence of the warning target, a historical stay duration in the stay record is acquired, and based on the historical stay duration, the whereabouts of the warning target is determined.

10. An electronic device, characterized in that: The electronic device includes a memory and a processor coupled to the memory, the memory stores at least one computer program, and when the at least one computer program is loaded and executed by the processor, it is used to implement the target tracking method as described in any one of claims 1-9.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium has at least one program, and when the at least one program is loaded and executed by the processor, it is used to implement the target tracking method according to any one of claims 1 to 9.

Images