Target object query method, electronic device and computer readable storage medium

Abstract

This application discloses a method for querying target objects, an electronic device, and a computer-readable storage medium. The method includes: obtaining query information corresponding to the target object; querying a database using the query information to obtain an initial recall result for the target object; in response to the number of initial recall results being less than a preset threshold, obtaining candidate query information matching the query information; performing at least one round of queries in the database using the query information and candidate query information to obtain candidate recall results for the target object; wherein the candidate recall results are determined based on the similarity between the query information and candidate query information and source information within the query range in the database, the query range changes with each round, and the similarity threshold matched in each round increases with each round; and determining the target recall result for the target object based on the initial recall result and the candidate recall result. This approach can improve the accuracy of target object queries.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method for querying target objects, an electronic device, and a computer-readable storage medium. Background Technology

[0002] With the rapid development and widespread adoption of mobile internet technology, as well as the advancement of transportation, a massive amount of video images has been generated. Currently, target objects are typically identified based on these video images, but issues such as shooting angle and image clarity result in low accuracy in target object retrieval. Therefore, improving the accuracy of target object retrieval has become an urgent problem to be solved. Summary of the Invention

[0003] The main technical problem addressed by this application is to provide a target object query method, electronic device, and computer-readable storage medium that can improve the accuracy of target object queries.

[0004] To address the aforementioned technical problems, this application provides a method for querying a target object, comprising: obtaining query information corresponding to the target object; querying a database using the query information to obtain an initial recall result for the target object; in response to the number of initial recall results being less than a preset threshold, obtaining candidate query information matching the query information; querying the database at least once using the query information and the candidate query information to obtain a candidate recall result for the target object; wherein the candidate recall result is determined based on the similarity between the query information and the candidate query information and source information within a query range in the database, the query range changing with each round, and the similarity threshold matched in each round increasing with each round; and determining a target recall result for the target object based on the initial recall result and the candidate recall result.

[0005] To address the aforementioned technical problems, a second aspect of this application provides an electronic device including a memory and a processor coupled to each other, wherein the memory stores program instructions and the processor executes the program instructions to implement the method described in the first aspect.

[0006] To address the aforementioned technical problems, a third aspect of this application provides a computer-readable storage medium storing program instructions executable by a processor, the program instructions being used to implement the method described in the first aspect.

[0007] The above scheme obtains the query information corresponding to the target object and uses the obtained query information to perform precise matching from the database to obtain the initial recall results corresponding to the target object. This ensures that the initial recall results are highly consistent with the query information. When the number of initial recall results is less than a preset threshold, candidate query information that matches the query information is obtained. The query information and candidate query information are used to perform at least one round of querying from the database to obtain candidate recall results for the target object. This avoids the inability to meet the requirements due to insufficient initial recall results. The candidate recall results are determined based on the similarity between the query information and candidate query information and the source information within the query range in the database. The query range changes with each round, and the similarity threshold matched in each round increases with each round to ensure that candidate recall results that are more closely related to the target object are selected. Based on the initial recall results and candidate recall results, the target recall results corresponding to the target object are obtained. Through multi-round progressive query retrieval and techniques such as increasing the similarity threshold with each round, the retrieval range is expanded, and the recall rate and precision rate of the target object are improved, thereby improving the query accuracy of the target object. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart illustrating one implementation method of the target object query method of this application; Figure 2 This is a flowchart illustrating another implementation of the target object query method of this application; Figure 3 This is a schematic diagram of the structure of one embodiment of the electronic device of this application; Figure 4 This is a schematic diagram of one embodiment of the computer-readable storage medium of this application. Detailed Implementation

[0009] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments, and different implementation methods can be adaptively combined. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0010] In this paper, the terms "system" and "network" are often used interchangeably. The term "and / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, "many" in this paper means two or more.

[0011] Please see Figure 1 , Figure 1 This is a flowchart illustrating one embodiment of the target object query method of this application. The method includes: S101: Obtain the query information corresponding to the target object, use the query information to query from the database, and obtain the initial recall result of the target object.

[0012] Specifically, the query information corresponding to the target object is obtained, and the obtained query information is used to perform precise matching from the database to obtain the initial recall result corresponding to the target object.

[0013] In one application approach, the system obtains the query information corresponding to the target object directly input by the user, extracts keywords or fields from the query information, and performs precise matching from the database based on the keywords or fields to obtain the initial recall results.

[0014] In another application, the system recommends multiple query items to the user, who can choose one as the information to be queried corresponding to the target object. The system then directly performs an accurate match from the database based on the information to be queried to obtain the initial recall results.

[0015] S102: In response to the number of initial recall results being less than a preset threshold, obtain candidate query information that matches the information to be queried, and use the information to be queried and the candidate query information to perform at least one round of querying from the database to obtain the candidate recall results of the target object.

[0016] Specifically, when the number of initial recall results is less than a preset threshold, candidate query information that matches the information to be queried is obtained, and at least one round of queries is performed from the database using the information to be queried and the candidate query information, thereby obtaining the candidate recall results of the target object.

[0017] The candidate recall results are determined based on the similarity between the information to be queried and the candidate query information and the source information within the query range in the database. The query range changes with each round, and the similarity threshold matched in each round increases with each round.

[0018] It should be noted that when the number of initial recall results is greater than or equal to the preset threshold, all initial recall results can be directly fed back to the user and the query can be stopped.

[0019] Understandably, the query scope will change with each round. For example, the query time range and query location range will change with each round, and the similarity threshold matched in each round will increase with each round. For example, the similarity threshold in the first round is 0.7 to ensure recall, and then increases by 0.05 in each subsequent round to improve accuracy.

[0020] Optionally, the preset quantity threshold can be 5, 7, or 10, etc., and can be set according to the actual situation. This application does not impose specific restrictions here.

[0021] In one application approach, common easily confused information is statistically analyzed based on historical data, and a misidentification database is constructed based on the easily confused information. Easily confused information in the information to be queried is identified, and corresponding candidate query information is obtained from the misidentification database. At least one round of queries is performed on the database using the information to be queried and the candidate query information, thereby obtaining the candidate recall results of the target object.

[0022] In another application, edit distance is used to find candidate query information similar to the query information, and the query information and candidate query information are used to perform at least one round of queries from the database to obtain candidate recall results of the target object.

[0023] S103: Based on the initial recall results and candidate recall results, determine the target recall result for the target object.

[0024] Specifically, based on the initial recall results and the candidate recall results, the target recall results corresponding to the target object are obtained.

[0025] In one application approach, the initial recall results and candidate recall results are merged and deduplicated to obtain the target recall results corresponding to the target object.

[0026] In another application, the initial recall results and candidate recall results are weighted according to a preset ratio, and a certain number are retained as the final target recall results.

[0027] The above scheme obtains the query information corresponding to the target object and uses the obtained query information to perform precise matching from the database to obtain the initial recall results corresponding to the target object. This ensures that the initial recall results are highly consistent with the query information. When the number of initial recall results is less than a preset threshold, candidate query information that matches the query information is obtained. The query information and candidate query information are used to perform at least one round of querying from the database to obtain candidate recall results for the target object. This avoids the inability to meet the requirements due to insufficient initial recall results. The candidate recall results are determined based on the similarity between the query information and candidate query information and the source information within the query range in the database. The query range changes with each round, and the similarity threshold matched in each round increases with each round to ensure that candidate recall results that are more closely related to the target object are selected. Based on the initial recall results and candidate recall results, the target recall results corresponding to the target object are obtained. Through multi-round progressive query retrieval and techniques such as increasing the similarity threshold with each round, the retrieval range is expanded, and the recall rate and precision rate of the target object are improved, thereby improving the query accuracy of the target object.

[0028] Please see Figure 2 , Figure 2 This is a flowchart illustrating another embodiment of the target object query method of this application. The method includes: S201: Obtain the query information corresponding to the target object, use the query information to query from the database, and obtain the initial recall result of the target object.

[0029] Specifically, the query information corresponding to the target object is obtained, and the obtained query information is used to perform precise matching from the database to obtain the initial recall result corresponding to the target object.

[0030] In one embodiment, the source information in the database includes target attributes and collection information corresponding to target objects collected from multiple collection locations. The information to be queried includes the target attributes of the target object, the collection time period, and the regional range. Step S201 specifically includes: obtaining target objects with consistent target attributes within the collection time period from the source information corresponding to the collection locations within the regional range, and obtaining the initial recall result; wherein, the initial recall result includes the target attributes of the target object, the collection time, and the collection location.

[0031] Specifically, from the source information corresponding to multiple collection locations within the region, target objects with consistent target attributes within the collection time period are accurately matched to obtain the initial recall results. Through the triple constraints of "regional scope", "collection time period" and "consistent target attributes", redundant information that does not meet the spatial, temporal and attribute conditions can be directly filtered out, ensuring that the initial results focus on the core target, thereby guaranteeing the accuracy and reliability of the initial recall results.

[0032] For ease of understanding, the following description takes a vehicle as the target object. It should be understood that the target object may also include other objects, such as animals.

[0033] In one application scenario, the information to be queried includes the target license plate number of the vehicle, the collection time period and the regional range. For example, the target license plate number is "Zhe ADD345", the collection time period is "08:00~10:00, June 26, 2025", and the regional range is "collection locations near City A". Through the target license plate number "Zhe ADD345", the vehicle with the license plate number Zhe ADD345 within 08:00~10:00, June 26, 2025 is obtained by exact matching from the source messages corresponding to multiple collection locations near City A, and other attribute information of the target license plate, such as license plate color, is obtained through analysis.

[0034] In a specific application scenario, the initial recall results are shown in Table 1 below: Table 1

[0035] In one implementation scenario, the database is obtained through the following steps: acquiring the target attributes and collection time corresponding to target objects collected by all collection locations, performing structured processing on the target attributes and collection time to obtain structured source information; performing encoding processing on the collection locations to obtain corresponding encoding values, establishing a mapping relation table between collection locations and encoding values, and obtaining the database based on the source information and the mapping relation table; wherein the regional range is determined based on the encoding values and the mapping relation table.

[0036] Specifically, acquiring the target attributes and collection time corresponding to target objects collected by all collection locations, and performing structured processing on the target attributes and collection time to obtain structured source information can avoid format confusion caused by different collection methods and recording methods of original data, ensure standardized data format in the database, and provide a unified standard for subsequent query; performing encoding processing on all collection locations to obtain corresponding encoding values can eliminate the problems of duplication, ambiguity or inconsistent expression of location names, and ensure that data references are unique, and establishing a mapping relation table between collection locations and encoding values can avoid data association failure caused by changes in collection location information. Finally, the database is automatically constructed based on the source information and the mapping relation table, thereby improving data standardization and consistency, reducing information confusion, realizing efficient associated integration of data, and improving the efficiency of subsequent target query.

[0037] In a specific implementation scenario, the data collected at the collection locations includes, but is not limited to: license plate number, collection time, collection location ID, collection latitude and longitude, and vehicle attributes (such as vehicle body color, license plate number, license plate color, vehicle model, etc.). The collected data undergoes structured processing, such as removing spaces from license plate numbers and standardizing letter case, to obtain structured source information. The range of collected data can be data from all target objects collected from all collection locations within a specific city within a single day. All collection locations are Geohash encoded to obtain corresponding Geohash values, and a mapping table between collection locations and Geohash values ​​is established. Based on the source information and the mapping table, a database is constructed. The regional range is determined by the Geohash values ​​and the mapping table; for example, the first four digits of the Geohash value correspond to a range of approximately 20km from the collection location, the first five digits correspond to a range of approximately 5km, and the complete six digits correspond to a range of approximately 1km-2km.

[0038] It should be noted that when new collected data is acquired, the old database is saved as historical data for reference, while the new collected data is restructured and encoded to automatically create a new database. For example, a database created using data on all target objects collected from all collection locations in a certain city yesterday will be saved. When it is necessary to analyze target objects appearing at all collection locations in another city today, the new collected data will be restructured and encoded to automatically create a new database. The new database can then be used to query target objects, and historical data from the old database can be retrieved for reference analysis when needed, thereby improving query efficiency and accuracy.

[0039] S202: In response to the number of initial recall results being less than a preset threshold, the target attribute in the query information is replaced with a candidate attribute to obtain candidate query information; wherein, the candidate attribute includes attributes whose similarity to the target attribute exceeds a preset threshold.

[0040] Specifically, when the number of initial recall results is less than a preset threshold, the target attribute in the query information is replaced with a candidate attribute to obtain candidate query information. The candidate attributes include attributes whose similarity to the target attribute exceeds a preset threshold, thereby expanding the query scope, reducing missed detections, and improving the accuracy of query results.

[0041] In an implementation scenario, the target object comprises a vehicle and the target attribute comprises a license plate number. In step S202, replacing the target attribute in the information to be queried with a candidate attribute to obtain candidate query information specifically includes: acquiring a target character in the license plate number, and obtaining a candidate character whose similarity with the target character exceeds a preset threshold based on the target character; replacing the target character with the candidate character to obtain a candidate license plate number, and taking the candidate license plate number as the candidate query information.

[0042] Specifically, a license plate number includes a plurality of visually similar characters, for example, D and 0, B and 8, etc. The target character in the license plate number can be replaced with a visually similar candidate character to obtain a candidate license plate number, and the candidate license plate number is used as candidate query information, so that the query range can be expanded, missed detection can be reduced, and the accuracy of the target vehicle query result can be improved.

[0043] In a specific implementation scenario, for example, if the license plate number is "Zhe A DD345", the target character "DD" therein can be replaced with the candidate character "00", so as to obtain the candidate license plate number "Zhe A 00345", and "Zhe A 00345" is used as the candidate query information.

[0044] S203: determining the query range of the current round based on the acquisition time period and the region range, and acquiring the similarity threshold of the current round.

[0045] Specifically, the query range of the current round is determined according to the acquisition time period and the region range, and the similarity threshold of the current round is acquired, wherein if the current round is the first round, the similarity threshold needs to be initialized.

[0046] S204: acquiring the similarity between the information to be queried, the candidate query information and source information within the query range, and determining a candidate recall result of the current round based on the similarity and the similarity threshold of the current round.

[0047] Specifically, the similarity between the information to be queried and the source information within the query range is acquired, and the similarity between the candidate query information and the source information within the query range is acquired; when the similarity is greater than the similarity threshold matched by the current round, the candidate recall result of the current round is obtained.

[0048] In one implementation scenario, step S204, obtaining the similarity between the information to be queried and the candidate query information and the source information within the query range, specifically includes: obtaining the target attribute features corresponding to the target attribute in the information to be queried, as well as the spatiotemporal features corresponding to the collection time period and area range; obtaining the candidate attribute features corresponding to the candidate attributes in the candidate query information, as well as the spatiotemporal features corresponding to the collection time period and area range; and obtaining the source attribute features corresponding to the target attribute in the source information within the query range, as well as the spatiotemporal features corresponding to the collection location and collection time. Based on the target attribute features and source attribute features, and the candidate attribute features and source attribute features, attribute feature similarity is obtained. Based on the spatiotemporal features corresponding to the information to be queried and the spatiotemporal features corresponding to the source information, and the spatiotemporal features corresponding to the candidate query information and the source information, spatiotemporal feature similarity is obtained. Based on attribute feature similarity and spatiotemporal feature similarity, the similarity between the information to be queried and the candidate query information and the source information within the query range is obtained.

[0049] Specifically, the target attribute features corresponding to the target attribute in the information to be queried include color attribute features and structural attribute features. The candidate attribute features corresponding to the candidate attributes in the candidate query information include color attribute features and structural attribute features. The source attribute features corresponding to the target attribute in the source information include color attribute features and structural attribute features. Based on the color attribute features, the corresponding color attribute feature similarity is obtained. Based on the structural attribute features, the corresponding structural attribute similarity is obtained. Based on the spatiotemporal features, the corresponding spatiotemporal feature similarity is obtained. Based on the color attribute feature similarity, structural attribute feature similarity, and spatiotemporal feature similarity, as well as the pre-assigned corresponding weight values, the similarity between the information to be queried and the candidate query information and the source information within the query range is calculated. By complementing multiple dimensions, the limitations of a single feature are reduced. By comprehensively analyzing the multi-dimensional similarity, the risk of mismatch can be reduced.

[0050] In one implementation scenario, before obtaining the similarity between the information to be queried and the candidate query information and the source information within the query range in step S204, the method further includes: obtaining the speed of the target object in the source information based on the collection time period and regional range in the information to be queried and the candidate query information, as well as the collection time and collection location in the source information within the query range; and removing the corresponding source information in response to the speed of the target object being greater than a preset speed threshold.

[0051] Specifically, based on the collection time period and area range in the information to be queried, as well as the collection time and location in the source information within the query range, the speed of the target object in the source information is obtained. When the speed of the target object is greater than a preset speed threshold, the corresponding source information is removed, thereby improving query efficiency.

[0052] To better understand the above solution, let's illustrate it with an example: In a specific implementation scenario, taking the target license plate number "Zhe A DD345" collected at Point 1 of City A at 8:00 a.m. as an example, a candidate license plate number "Zhe A 00345" is generated first, then license plate numbers with similar structures, for example, with an edit distance less than or equal to 2 are obtained. Next, the collection location is quickly located to nearby points within the range of 1km-2km of City A through the complete 6-digit Geohash value, and a fuzzy query is performed within a small time range, for example, between 7:50 and 8:20, and the results shown in Table 2 below are obtained.

[0053] Table 2

[0054] Further, the speed of the target object in each matched source information is calculated first. Among them, the vehicle with the license plate number "Zhe A 00335" appeared at Point 1 of the Exit of C Road Subway at 8:19, while the vehicle with the target license plate number "Zhe A DD345" appeared at Point 3 of City A at 8:20 in the initial recall result, and the distance between City A and C Road Subway Exit is 1.8km, so the calculated speed is 108km / h. Since the speed limit on urban roads is 40-60km / h, this result is unreasonable and needs to be removed. Then the similarity between the information to be queried, candidate query information and the source information within the query range is calculated, and the obtained results are shown in Table 3 below.

[0055] Table 3

[0056] Wherein, the weight corresponding to the spatio-temporal feature is 0.6, the weight corresponding to the structural attribute feature is 0.3, and the weight corresponding to the color attribute feature is 0.1. The preset speed threshold is determined according to road properties: if it is an urban road, the preset speed threshold is 40-60km / h; if it is an expressway, the preset speed threshold is 80-120km / h. The calculation formula for the similarity of structural attribute features = 1 - edit distance from the target license plate / length of the target license plate. The license plate color similarity = 1 if the color is consistent with the target license plate, otherwise it is 0.4. The current round is the first round, and the corresponding similarity threshold is 0.7. Therefore, Zhe A DD445, Zhe A DC145 and Zhe A 00445 are retained, and Zhe A 00335 is removed to obtain the candidate recall result of the current round.

[0057] S205: Based on the collection time and collection location of the target object in the candidate recall result, update the collection time period and area range, and update the similarity threshold for the next round, until the iteration termination condition is satisfied.

[0058] Specifically, based on the collection time and location of the target objects in the candidate recall results, the collection time period and area range are continuously updated. That is, the candidate recall results of this round are used as candidate query information for the next round, and the similarity threshold for the next round is updated. For example, the similarity threshold for the next round is 0.75, until the iteration termination condition is met. The iteration termination condition is related to the number of candidate recall results and / or the number of iteration rounds. For example, if the number of rounds has reached 5, the iteration will be terminated, or the iteration will also be terminated when the number of new candidate recalls is less than 2.

[0059] S206: Based on the initial recall results and candidate recall results, determine the target recall result for the target object.

[0060] Specifically, based on the initial recall results and all candidate recall results, the target recall results for the target object are obtained.

[0061] In a specific implementation scenario, the target recall results are shown in Table 4 below.

[0062] Table 4

[0063] S207: Based on the collection time corresponding to the target object in the target recall results, sort the target recall results; based on the collection location of the target object in the sorted target recall results, obtain the trajectory of the target object.

[0064] Specifically, the target recall results are sorted according to the collection time corresponding to the target object in the target recall results, and the trajectory of the target object is obtained according to the collection location of the target object in the sorted target recall results, thereby solving the problem of reconstructing the trajectory of the target object.

[0065] The above scheme obtains the query information corresponding to the target object and uses the obtained query information to perform precise matching from the database to obtain the initial recall results corresponding to the target object. When the number of initial recall results is less than a preset threshold, the target attribute in the query information is replaced with a candidate attribute to obtain candidate query information. Based on the collection time period and regional range, the query range of the current round is determined, and the similarity threshold of the current round is obtained. The similarity between the query information and the source information within the query range, as well as the similarity between the candidate query information and the source information within the query range, are obtained. When the similarity is greater than the similarity threshold of the current round, the candidate recall results of the current round are obtained. Based on the collection time and collection location of the target object in the candidate recall results, the collection time period and regional range are updated again, that is, this round... The candidate recall results are used as candidate query information for the next round, and the similarity threshold for the next round is updated. For example, the similarity threshold for the next round is 0.75, until the iteration termination condition is met. Based on the initial recall results and all candidate recall results, the target recall results of the target object are obtained. The target recall results are sorted according to the collection time corresponding to the target object in the target recall results, and the trajectory of the target object is obtained according to the collection location of the target object in the sorted target recall results. By using round counters and convergence detection as iteration termination conditions, and combining spatiotemporal correlation constraints and dynamic similarity adjustment, the retrieval scope is expanded while avoiding invalid expansion, thus improving retrieval efficiency. Furthermore, the chain retrieval improves the query accuracy of the target object, thereby solving the problem of reconstructing the trajectory of the target object.

[0066] Please see Figure 3 , Figure 3This is a schematic diagram of an embodiment of the electronic device of this application. The electronic device 30 includes a memory 300 and a processor 302 coupled to each other. The memory 300 stores program data (not shown). The processor 302 calls the program data to implement the method in any of the above embodiments. For related descriptions, please refer to the detailed description of the above method embodiments, which will not be repeated here. Specifically, the electronic device 30 includes: desktop computers, laptops, tablet computers, servers, etc., which are not limited here. In addition, the processor 302 can also be called a CPU (Center Processing Unit). The processor 302 may be an integrated circuit chip with signal processing capabilities. The processor 302 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor. In addition, the processor 302 can be implemented by integrated circuit chips.

[0067] Please see Figure 4 , Figure 4 This is a schematic diagram of a computer-readable storage medium according to an embodiment of the present application. The computer-readable storage medium 40 stores program data 400. When the program data 400 is executed by a processor, it implements the method in any of the above embodiments. For a detailed description of the relevant content, please refer to the detailed description of the above method embodiments, which will not be repeated here.

[0068] It should be noted that the units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0069] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0070] If the integrated unit is implemented as 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 this application, 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. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0071] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for querying target objects, characterized in that, include: Obtain the query information corresponding to the target object, and use the query information to query from the database to obtain the initial recall result of the target object; In response to the initial recall result being less than a preset threshold, candidate query information matching the information to be queried is obtained. At least one round of querying is performed on the database using the information to be queried and the candidate query information to obtain candidate recall results for the target object. The candidate recall results are determined based on the similarity between the information to be queried and the candidate query information and the source information within the query range in the database. The query range changes with each round, and the similarity threshold matched in each round increases with each round. The source information in the database includes target attributes and collection times corresponding to target objects collected from multiple collection locations. The information to be queried includes the target attributes, collection time period, and regional range of the target object. The query range for the current round is determined based on the collection time period and the regional range, and the collection time period and the regional range are updated based on the collection time and collection location of the target object in the candidate recall results. Based on the initial recall results and the candidate recall results, the target recall result for the target object is determined.

2. The method according to claim 1, characterized in that, The step of querying the database using the information to be queried to obtain the initial recall result of the target object includes: From the source information corresponding to the collection location within the area, obtain the target object with the same target attribute within the collection time period to obtain the initial recall result; wherein, the initial recall result includes the target attribute of the target object, the collection time and collection location.

3. The method according to claim 2, characterized in that, The database was obtained based on the following steps: Obtain the target attributes and collection time corresponding to the target objects collected at all the collection locations, and perform structured processing on the target attributes and collection time to obtain the structured source information; The collection locations are encoded to obtain corresponding encoded values, and a mapping table between the collection locations and the encoded values ​​is established. Based on the source information and the mapping table, the database is obtained; wherein, the area range is determined based on the encoded values ​​and the mapping table.

4. The method according to claim 2, characterized in that, The response that the number of initial recall results is less than a preset threshold is used to obtain candidate query information matching the information to be queried, and to perform at least one round of querying from the database using the information to be queried and the candidate query information to obtain candidate recall results for the target object, including: In response to the number of initial recall results being less than the preset number threshold, the target attribute in the query information is replaced with a candidate attribute to obtain the candidate query information; wherein, the candidate attribute includes attributes whose similarity to the target attribute exceeds a preset threshold; Based on the collection time period and the area range, the query range for the current round is determined, and the similarity threshold for the current round is obtained. Obtain the similarity between the information to be queried and the candidate query information and the source information within the query range, and determine the candidate recall result for the current round based on the similarity and the similarity threshold of the current round; Based on the collection time and location of the target objects in the candidate recall results, the collection time period and the regional range are updated, and the similarity threshold for the next round is updated until the iteration termination condition is met; wherein, the iteration termination condition is related to the number of candidate recall results and / or the number of iteration rounds.

5. The method according to claim 4, characterized in that, The step of obtaining the similarity between the information to be queried and the candidate query information and the source information within the query range includes: Obtain the target attribute features corresponding to the target attribute in the information to be queried, as well as the spatiotemporal features corresponding to the collection time period and the area range; obtain the candidate attribute features corresponding to the candidate attributes in the candidate query information, as well as the spatiotemporal features corresponding to the collection time period and the area range; and obtain the source attribute features corresponding to the target attribute in the source information within the query range, as well as the spatiotemporal features corresponding to the collection location and the collection time. Based on the target attribute features and the source attribute features, as well as the candidate attribute features and the source attribute features, the attribute feature similarity is obtained. Based on the spatiotemporal features corresponding to the information to be queried and the spatiotemporal features corresponding to the source information, as well as the spatiotemporal features corresponding to the candidate query information and the spatiotemporal features corresponding to the source information, the spatiotemporal feature similarity is obtained. Based on the attribute feature similarity and the spatiotemporal feature similarity, the similarity between the information to be queried and the candidate query information and the source information within the query range is obtained.

6. The method according to claim 5, characterized in that, Before obtaining the similarity between the information to be queried and the candidate query information and the source information within the query range, the method further includes: Based on the collection time period and the regional range in the information to be queried and the candidate query information, as well as the collection time and the collection location in the source information within the query range, the speed of the target object in the source information is obtained; In response to the target object's speed exceeding a preset speed threshold, the corresponding source information is removed.

7. The method according to claim 4, characterized in that, The target object includes the target vehicle, and the target attribute includes the license plate number. The step of replacing the target attribute in the information to be queried with candidate attributes to obtain candidate query information includes: Obtain the target character from the license plate number, and based on the target character, obtain candidate characters whose similarity to the target character exceeds the preset threshold; The target character is replaced with the candidate character to obtain the candidate license plate number, and the candidate license plate number is used as the candidate query information.

8. The method according to claim 4, characterized in that, After determining the target recall result of the target object based on the initial recall result and the candidate recall result, the process includes: Based on the collection time corresponding to the target object in the target recall results, the target recall results are sorted. Based on the collection locations of the target objects in the sorted target recall results, the trajectory of the target objects is obtained.

9. An electronic device, characterized in that, The method includes a memory and a processor coupled to each other, the memory storing program instructions, and the processor executing the program instructions to implement the method according to any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The system stores program instructions that can be executed by a processor, the program instructions being used to implement the method according to any one of claims 1-8.

Images