Control management method and device, platform and storage medium

Abstract

The application relates to the technical field of computers and discloses a deployment control management method, device, platform and storage medium, the method comprising the following steps: obtaining a target deployment control task and task information of the target deployment control task, wherein the task information comprises a target deployment control area and a target deployment control object; determining a target deployment control system from accessed deployment control systems based on the target deployment control area; if the target deployment control area is greater than a preset range, obtaining a large-database deployment control task corresponding to the target deployment control task in the target deployment control system, wherein the large-database deployment control task is used for controlling a plurality of deployment control objects in the same target deployment control area; and writing the target deployment control object into the large-database deployment control task to control the target deployment control object and obtain a deployment control result. The method can solve the problems of complex deployment control process and low deployment control efficiency when a large deployment control area is set with a deployment control task.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and more specifically to deployment and management methods, devices, platforms, and storage media. Background Technology

[0002] With the increasing sophistication of video surveillance infrastructure, facial images can be captured in designated surveillance areas using video surveillance equipment. These images are then transmitted over a network to the surveillance system's backend for identification and comparison, accurately identifying monitored individuals and achieving better monitoring results. When deploying surveillance, the person in charge needs to fill in specific deployment tasks on the surveillance management platform, including information such as the surveillance name, deployment action, start time, end time, and area. However, for large-scale surveillance tasks, the large area involved and the amount of information required complicate the deployment process and reduce efficiency. Summary of the Invention

[0003] In view of this, the present invention provides a deployment management method, device, platform and storage medium to solve the problems of complex deployment process and low deployment efficiency when setting deployment tasks for large deployment areas.

[0004] In a first aspect, the present invention provides a deployment management method, the method comprising:

[0005] Acquire the target deployment task and its task information, wherein the task information includes the target deployment area and the target deployment object;

[0006] Based on the target control area, the target control system is determined from the connected control systems;

[0007] If the target control area is larger than a preset range, obtain the large warehouse control task corresponding to the target control task in the target control system. The large warehouse control task is used to control multiple control objects in the same target control area.

[0008] The target object is written into the large database control task to control the target object and obtain the control result.

[0009] In this approach, a large-scale warehouse control task is deployed within the target control system. Therefore, when the target control area of ​​the acquired target control task is larger than a preset range, a large-scale warehouse control task corresponding to the target control task can be obtained, and the target control object can be directly written into the large-scale warehouse control task to control the target control object. Thus, when the control area is large, the task information filling process and the control process can be simplified, thereby improving control efficiency.

[0010] In an optional implementation, the method further includes:

[0011] If the target control area is less than or equal to the preset range, the target control task and its target control object are sent to the target control system to control the target control object and obtain the control result.

[0012] In this method, when the target control area is less than or equal to the preset range, the target control task and its target control object are directly sent to the target control system, thus ensuring the control efficiency of small-scale target control tasks.

[0013] In an optional implementation, the method further includes:

[0014] When a target deployment result is received from any of the connected deployment systems, the target task code corresponding to the target deployment result is obtained;

[0015] When the target task is encoded as the task code of the large database deployment task, the deployment object corresponding to the target deployment result is obtained;

[0016] Based on the control object corresponding to the target control result, determine the control task and its status corresponding to the target control result from the pre-stored control tasks corresponding to the large warehouse control task;

[0017] When the task status meets the preset effective deployment conditions, the target deployment result is sent to the corresponding business system according to the deployment task corresponding to the target deployment result.

[0018] In this approach, when the received target deployment result is a deployment result fed back from a large database deployment task, the corresponding deployment task and its status are first queried based on the deployment object corresponding to the target deployment result. If the task status meets the preset valid deployment conditions, the target deployment result is then sent to the corresponding business system. Therefore, it avoids feeding back invalid deployment results to the corresponding business system, thereby reducing unnecessary storage and data transmission resource consumption and improving deployment efficiency.

[0019] In an optional implementation, the method further includes:

[0020] When the target task code is not the task code of the large database deployment task, the deployment task corresponding to the target deployment result is determined according to the target task code, so as to send the target deployment result to the corresponding business system.

[0021] In this approach, when the received target deployment result is not the deployment result of the large database deployment task, the target deployment result is directly sent to the corresponding business system according to the target task code, thus ensuring the transmission efficiency of the deployment result of the small-scale deployment task.

[0022] In an optional implementation, the method further includes:

[0023] Obtain a preset system assessment task, wherein the system assessment task includes at least one assessment sub-task;

[0024] The accessed deployment system is evaluated based on the evaluation sub-tasks to obtain the task evaluation results of each accessed deployment system corresponding to each evaluation sub-task.

[0025] For each connected deployment system, the task assessment results of all the assessment sub-tasks are merged to obtain the system assessment result of the connected deployment system.

[0026] When the system assessment result indicates that the control system fails to meet the standards, a corresponding system alarm message is generated.

[0027] In this approach, each connected control system is assessed through a pre-set system assessment task. When the system assessment result of a control system is found to be substandard, a corresponding system alarm message is generated. Therefore, relevant personnel can promptly check the control system to ensure the effective execution of the control of the target task, thereby improving control efficiency and quality.

[0028] In one optional implementation, the step of evaluating the accessed deployment systems based on the evaluation sub-tasks to obtain the task evaluation results for each accessed deployment system corresponding to each evaluation sub-task includes:

[0029] The assessment sub-tasks are analyzed to determine the task objectives;

[0030] Based on the task objective, obtain the processing result samples and target processing results corresponding to each connected deployment system;

[0031] Based on the target processing results of each connected deployment system and the corresponding processing result samples, the task assessment results of each connected deployment system corresponding to the assessment sub-task are obtained.

[0032] In this approach, before evaluating each deployment system, the task objective of the current evaluation sub-task is first determined. Then, based on the task objective, the processing result samples and target processing results corresponding to each deployment system are obtained and compared to obtain the task evaluation results of the deployment system. Therefore, it is possible to effectively determine whether each deployment system meets the corresponding evaluation requirements and whether the performance of the deployment system is stable.

[0033] In one optional implementation, the step of obtaining processing result samples corresponding to each accessed deployment system and the target processing result based on the task objective includes:

[0034] If the task objective is to assess computational performance, computational control samples corresponding to each connected control system are obtained from the assessment sub-tasks; each computational control sample is sent as a control object to the corresponding control system, so that the corresponding control system controls the computational control sample to obtain the processing result sample corresponding to each connected control system; each processing result sample is sent as a control object to the corresponding control system, so that the corresponding control system controls the processing result sample to obtain the target processing result corresponding to each connected control system.

[0035] Alternatively, if the task objective is to assess the performance of image search, an external image quality detection interface is invoked to perform quality detection on the historical images captured in the deployment results fed back by each connected deployment system, thereby obtaining a first quality score for the historical images; based on the first quality score of the historical images, historical images with the first quality score falling within a first preset score range are selected from the historical images corresponding to each connected deployment system, as processing result samples corresponding to each connected deployment system; each processing result sample is then sent to the corresponding deployment system for image search to obtain the target processing result corresponding to each connected deployment system;

[0036] Alternatively, if the task objective is recall rate assessment, the control objects with trajectories within the preset recall rate assessment days are determined from the control results fed back by each connected control system; the control objects with trajectories are sent to the corresponding control system for image search to obtain the processing result samples corresponding to each connected control system; the trajectories of the control objects with trajectories under the same time and the same device are obtained from the corresponding control system as the target processing result;

[0037] Alternatively, if the task objective is accuracy assessment, the system obtains control objects with a preset number of trajectories within a preset number of accuracy assessment days from the local system to obtain processing result samples corresponding to each connected control system; from the corresponding control system, the system obtains all trajectories of the processing result samples to obtain the target processing result corresponding to each connected control system.

[0038] Alternatively, if the task objective is to assess the pass rate of image-based document search, an external image quality detection interface is invoked to perform quality detection on the cover photos of the archives of the monitored objects corresponding to each connected local control system, thereby obtaining a second quality score for the cover photos. Based on the second quality score of the cover photos, cover photos with the second quality score falling within a second preset score range are selected from the cover photos corresponding to each connected control system to obtain processing result samples corresponding to each connected control system. Each processing result sample is then sent to the corresponding control system for image-based document search to obtain the target processing result corresponding to each connected control system.

[0039] In this approach, for different task objectives, the processing result sample of the current task objective is determined from the data fed back from the local system or the control system to be evaluated, and then the target processing result corresponding to the processing result sample is obtained from the control system to be evaluated. Therefore, the performance of each control system connected under the current task objective can be effectively evaluated by using the processing result sample and target processing result corresponding to each connected control system.

[0040] Secondly, the present invention provides a deployment management device, the device comprising:

[0041] The deployment task acquisition module is used to acquire target deployment tasks and their task information, wherein the task information includes target deployment area and target deployment object;

[0042] The deployment system selection module is used to determine the target deployment system from the connected deployment systems based on the target deployment area;

[0043] The large warehouse task selection module is used to obtain the large warehouse deployment task corresponding to the target deployment task in the target deployment system if the target deployment area is larger than a preset range. The large warehouse deployment task is used to deploy multiple deployment objects in the same target deployment area.

[0044] The target control module is used to write the target control object into the large database control task in order to control the target control object and obtain the control result.

[0045] Thirdly, the present invention provides a deployment management platform, comprising:

[0046] A view library is used to issue target deployment tasks and their task information, the task information including the target deployment area and the target deployment object;

[0047] A multi-algorithm management and scheduling engine system, which is connected to the view library, is used to execute the deployment management method of the first aspect or any corresponding embodiment described above;

[0048] At least one deployment system is connected to the multi-algorithm management and scheduling engine system, and is used to deploy the target deployment object under the scheduling of the multi-algorithm management and scheduling engine system to obtain deployment results.

[0049] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the deployment and management method of the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0050] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0051] Figure 1 This is a flowchart illustrating the first deployment and management method according to an embodiment of the present invention;

[0052] Figure 2 This is a flowchart illustrating the second deployment and management method according to an embodiment of the present invention;

[0053] Figure 3 This is a framework diagram of a multi-engine parsing platform according to an embodiment of the present invention.

[0054] Figure 4 This is a schematic diagram of the data interaction logic of a multi-engine parsing platform in a large-scale deployment scenario according to an embodiment of the present invention;

[0055] Figure 5 This is a schematic diagram of the overall data interaction logic of the multi-engine parsing platform according to an embodiment of the present invention;

[0056] Figure 6 This is a flowchart illustrating the third deployment and management method according to an embodiment of the present invention;

[0057] Figure 7 This is a schematic diagram of the deployment process of the deployment object according to an embodiment of the present invention;

[0058] Figure 8 This is a schematic diagram of the data aggregation service according to an embodiment of the present invention;

[0059] Figure 9 This is a schematic diagram of the image search service according to an embodiment of the present invention;

[0060] Figure 10This is a structural block diagram of a deployment management device according to an embodiment of the present invention. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] With the increasing sophistication of video surveillance infrastructure, facial images can be captured in designated surveillance areas using video surveillance equipment. These images are then transmitted over a network to the surveillance system's backend for identification and comparison, accurately identifying monitored individuals and achieving better monitoring results. However, when assigning surveillance tasks to individuals, the person in charge needs to fill out specific task information on the surveillance management platform. For large-scale surveillance tasks, the large area involved and the amount of information required complicate the process and reduce efficiency.

[0063] In view of this, according to an embodiment of the present invention, a deployment management method embodiment is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0064] This embodiment provides a deployment management method that can be used to access deployment management platforms of different deployment systems. Figure 1 This is a flowchart of a deployment and management method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps:

[0065] Step S101: Obtain the target deployment task and its task information, wherein the task information includes the target deployment area and the target deployment object.

[0066] It should be noted that, depending on the actual situation, the task information may also include the task code, task type, and object code of the target control object. If each connected control system already has multiple large-scale control tasks, it is not necessary to create a new target control task; instead, the target control object can be parsed and placed into the corresponding control task.

[0067] Step S102: Based on the target control area, determine the target control system from the connected control systems.

[0068] Step S103: If the target control area is larger than a preset range, obtain the large warehouse control task corresponding to the target control task in the target control system. The large warehouse control task is used to control multiple control objects in the same target control area.

[0069] It should be noted that before obtaining the large-database deployment task corresponding to the target deployment task in the target deployment system, an instruction to create a large-database deployment task can be sent to each deployment system offline, so that each deployment system can create the large-database deployment task. The large-database deployment task has the following attribute requirements: the large-database deployment task is always valid and only needs to be created once; the task code is fixed; the deployment area is larger than a preset range; the deployment threshold is fixed (e.g., 95%); the deployment time is always valid; the task deployment status is always "deploying"; and the task code of the large-database deployment task must be provided to the task issuing object (e.g., the view library). Of course, in actual practice, multiple large-database deployment tasks can be created according to actual needs, and the number is not limited here.

[0070] It should be noted that the preset range is determined based on the actual needs of the corresponding large-scale warehouse deployment task.

[0071] Step S104: Write the target control object into the large database control task to control the target control object and obtain the control result.

[0072] It should be noted that each deployment system has a Kafka task-based alarm topic queue. The deployment results of the deployment system based on the large database deployment task can be put into the Kafka task-based alarm topic queue for reading.

[0073] The deployment management method provided in this embodiment deploys large-scale warehouse deployment tasks in the target deployment system. Therefore, when the target deployment area of ​​the acquired target deployment task is larger than a preset range, a large-scale warehouse deployment task corresponding to the target deployment task can be acquired, and the target deployment object can be directly written into the large-scale warehouse deployment task to deploy the target deployment object. Thus, when the deployment area is large, the task information filling process and deployment process can be simplified, thereby improving deployment efficiency.

[0074] Figure 2 This is a flowchart of a deployment and management method according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps:

[0075] Step S201: Obtain the target deployment task and its task information, including the target deployment area and the target deployment object. See step S101 above for details, which will not be elaborated further here.

[0076] Step S202: Based on the target control area, determine the target control system from the connected control systems. See step S102 above for details, which will not be elaborated further here.

[0077] Step S203: If the target control area is larger than a preset range, obtain the large-database control task corresponding to the target control task in the target control system. The large-database control task is used to control multiple control objects within the same target control area. See step S103 above for details, which will not be elaborated further here.

[0078] Step S204: Write the target control object into the large database control task to control the target control object and obtain the control result.

[0079] For example, by assigning the corresponding task code of the large database deployment task to the target deployment object, the target deployment object is sent to the corresponding target deployment system in accordance with the large database deployment task method.

[0080] Furthermore, the method also includes:

[0081] Step S205: If the target control area is less than or equal to the preset range, the target control task and its target control object are sent to the target control system to control the target control object and obtain the control result.

[0082] It should be noted that in actual operation, in addition to determining the target deployment task to be directly sent to the target deployment system in the target deployment area based on the scope of the target deployment area, or writing the target deployment object of the target deployment task into the corresponding large database deployment task, the task publishing object (such as: view library) can also write the task type into the task information based on the scope of the target deployment area, and determine the deployment area scope corresponding to the current target deployment task based on the task type in the task information, so as to directly send its target deployment object to the target deployment system or write it into the large database deployment task and then send it to the target deployment system.

[0083] The deployment management method provided in this embodiment directly sends the target deployment task and its target deployment object to the target deployment system when the target deployment area is less than or equal to the preset range, thus ensuring the deployment efficiency of small-scale target deployment tasks.

[0084] Specifically, the method further includes:

[0085] Step S206: When a target deployment result is received from any of the connected deployment systems, the target task code corresponding to the target deployment result is obtained.

[0086] Step S207: When the target task code is the task code of the large database deployment task, obtain the deployment object corresponding to the target deployment result.

[0087] It should be noted that if the task code of the control task associated with the target control result is the task code of the large database control task, then it can be determined that the current target control result is based on the feedback of the large database control task. However, since the control object associated with the large database control task may correspond to different control tasks, it is necessary to further obtain the task object corresponding to the target control result in order to query the corresponding control task based on the task object corresponding to the target control result, and then send the target control result to the business system corresponding to the control task.

[0088] Step S208: Based on the control object corresponding to the target control result, determine the control task corresponding to the target control result and its task status from the pre-stored control tasks corresponding to the large warehouse control task.

[0089] Specifically, the object code of the controlled object associated with the target control result can be obtained. Based on the object code of the associated controlled object, the task code of the associated control task can be retrieved to obtain the task status of the associated control task. Then, based on the task status, it can be determined whether the target control task is under control. Specifically, the task status includes information such as the control status, control area, checkpoint, control equipment, and effective time of the control task corresponding to the target control result.

[0090] Step S209: When the task status meets the preset effective deployment conditions, the target deployment result is sent to the corresponding business system according to the deployment task corresponding to the target deployment result.

[0091] Specifically, based on the task status, it is determined whether the task is still under control, the task's effective time, the control area corresponding to the object code of the controlled object, and the control device meet the control task requirements. This determines whether the task status meets the valid control conditions, i.e., whether the control task of the target control result is valid, and thus whether the target control result is valid. It should be noted that if the target control task corresponding to the control result does not meet the valid control conditions, the target control result is determined to be invalid, and therefore it is not necessary to send the target control result to the corresponding business system.

[0092] The deployment management method provided in this embodiment, when the received target deployment result is the target deployment result fed back by the large database deployment task, first queries the corresponding deployment task and its status based on the deployment object corresponding to the target deployment result. If the task status meets the preset valid deployment conditions, then the target deployment result is sent to the corresponding business system. Therefore, it can avoid feeding back invalid deployment results to the corresponding business system, thereby reducing unnecessary storage and data transmission resource consumption and improving deployment efficiency.

[0093] Furthermore, the method also includes:

[0094] Step S210: When the target task code is not the task code of the large database deployment task, determine the deployment task corresponding to the target deployment result according to the target task code, so as to send the target deployment result to the corresponding business system.

[0095] Understandably, if the task code of the associated deployment task does not belong to the task code of the large database deployment task, it can be determined that the current target deployment result is not based on the feedback from the large database deployment task. Therefore, the target deployment result can be directly fed back to the business system corresponding to the task code of the deployment task.

[0096] The deployment management method provided in this embodiment can directly send the target deployment result to the corresponding business system according to the target task code when the received target deployment result is not the deployment result of the large database deployment task. Therefore, it can ensure the transmission efficiency of the deployment result of the small-scale deployment task.

[0097] For example, see Figure 3Taking the software architecture of the deployment management method of this invention, the "multi-engine parsing platform," as an example, the multi-engine parsing platform consists of two parts: an external system and an internal system. 1. External System: View library, video image service support platform, and various deployment systems; wherein, each deployment system corresponds to a deployment area and calls algorithms for different deployment areas based on task requirements. 2. Internal System: Multi-algorithm management and scheduling engine system, which includes: a message queue module, an algorithm supervision and evaluation module, an audit log module, a data reconciliation module, an operation and maintenance management module, an interface scheduling module, and an algorithm scheduling module. Specifically, the message queue module sends algorithm analysis results to Kafka, where the multi-engine module parses, consumes, and stores them, then forwards them to the view library. The algorithm supervision and evaluation module assesses whether the algorithms of each deployment system meet user requirements, whether their performance meets standards, and whether they are stable, and displays the assessment results in the operation and maintenance management module. Based on the data information displayed on the page, it determines whether the algorithm of the current deployment system meets the requirements. The audit log module is responsible for processing call information and details between various modules and displaying them to the operation and maintenance management module according to a certain pattern. The data reconciliation module compares the analysis data stored in the view library with the analysis data stored by the algorithm itself, and displays the differences to the operations and maintenance management module. The operations and maintenance management module receives data from various modules, displays the data, and manages the online status of the deployment system. The interface scheduling module and algorithm scheduling module handle tasks issued by the view library and third-party services, forwarding them to various deployment systems according to REST protocol messages, and then organizing the callback to the view library based on the returned results.

[0098] Specifically, the multi-algorithm management and scheduling engine system receives image processing tasks dispatched by the application, creates intelligent analysis tasks, intelligently parses various images, supports traffic load balancing, and pushes analysis tasks and results using standard protocols. Tasks are orchestrated and categorized for distribution between the application and the algorithms of multiple deployment systems, thereby shielding the algorithms of multiple deployment systems. Specifically, the view library can transmit and interact with the multi-algorithm management and scheduling engine platform via Kafka or an interface, selecting the transmission method based on the data volume. For example, if the task is a target deployment task, and this target deployment task is a large database deployment task in a large deployment area, then the data traffic of the deployment object is large and needs to be distributed to multiple deployment areas, then Kafka is used to transmit the data. If only a single deployment object is targeted, then an interface can be used to transmit the data.

[0099] Understandably, the multi-algorithm management and scheduling engine system, as a platform for managing various deployment systems, manages the resource utilization, functional availability, concurrency stability, and fault tolerance of each deployment system through assessment, monitoring, and alarms. In the aforementioned multi-algorithm management and scheduling engine system, service registration, configuration center, assessment center, and request routing gateway technologies (such as service registration, configuration center, and request routing gateway deployed on the Dolphin management platform) can be used to achieve one-click startup and deployment. Service registration is used to observe whether the deployment system is stable; the assessment center is used to assess whether it meets various functional, performance, and stability requirements of the customer within a certain period; and the routing gateway technology provides a unique entry point for the system, authenticating requests from requesters, identifying the permissions of each request, and intercepting abnormal requests, thus ensuring the security of the backend services.

[0100] For example, see Figure 4 and Figure 5 Taking a multi-engine parsing platform in a large-scale deployment scenario, with the target deployment task being the user's deployment task as an example, the deployment management method of the present invention will be further explained:

[0101] 1. The view library is responsible for storing captured data, and images of objects to be monitored and associated information can be uploaded to the view library. The associated information includes the target monitoring area. For example... Figure 4 As shown, users can pre-set monitoring periods based on various business needs or predefined rules, such as algorithm monitoring tasks. Within the monitoring period, algorithm data from each control system is aggregated. Upon reaching a specified time, the algorithm data is analyzed to monitor and evaluate each control system, triggering a task and distributing it to the multi-algorithm and scheduling engine system. Furthermore, if the task is a large-database control task, the view library identifies the target control task's task code (i.e., task ID) and task type as a large-database control task and distributes it to the multi-algorithm and scheduling engine system. It should be noted that the view library can first generate user control tasks and distribute them to the multi-algorithm and scheduling engine system. Users then allocate corresponding control objects based on the distributed tasks, diverting multiple control objects to multiple user control tasks or pre-built large-database control tasks based on control requirements. For example, if 100 control objects need to be controlled on a large scale, and multiple large-scale control tasks already exist, there is no need to create new tasks; the 100 control objects can be parsed and placed into the corresponding control tasks.

[0102] 2. The multi-algorithm management and scheduling engine system receives target deployment tasks issued by the view library. Figure 4(The text is incomplete and contains several errors. A more accurate translation would require the full context.) If the user deployment task is a small task targeting a single object, it can be sent to the corresponding target deployment area based on the task information, thus invoking the corresponding deployment system for processing. If the user deployment task targets a large area, the task information is recorded, and the target object is converted into a large-database deployment task object. The target object is then assigned the corresponding large-database deployment task's task code (shown as the large-database task ID in the diagram), and sent to the deployment system as a large-database deployment task. Furthermore, the multi-algorithm management and scheduling engine system, based on the received information about the deployed object in a large-database deployment task and the user deployment task's task code, can establish a large-database deployment task table and a deployment object memory table. This caches the user deployment task information and the target deployment object information in memory, enabling subsequent rapid calculation and comparison. Furthermore, instructions to create large database tasks were sent to each control system offline, allowing each control system to create large database control tasks.

[0103] 3. Deployment System: Each deployment system is responsible for analyzing and comparing captured data, receiving scheduling from multiple engine platforms, and returning the results to the view library for display. Specifically, the deployment system has a Kafka task-based alarm topic queue. Based on the received deployment tasks, the deployment system adds tasks with large-database deployment task codes to the large-database deployment database for deployment, and deploys small-database deployment tasks based on requirements. If alarm information is generated based on the deployment results, the alarm information is placed in the Kafka task-based alarm topic queue.

[0104] 4. The view library reads the alarm information of small control tasks in the Kafka task-type alarm topic queue, that is, the alarm information of control tasks in the message queue whose task code is not the task code of the large database control task, and sends the corresponding alarm information to the corresponding business system based on the task code of the control task in the read alarm information.

[0105] 5. The multi-algorithm management and scheduling engine system reads alarm information from the Kafka task-type alarm topic queue where the task code of the deployment task is the same as the task code of the large database deployment task. It obtains the object code of the associated deployment object (the deployment object ID in the diagram). By querying the large database deployment object memory table, it finds the task code of the actual user deployment task associated with the object code of this deployment object. Then, it queries the large database deployment task memory table to obtain information such as the deployment area, checkpoint, device, task validity time, and task status of the corresponding task. Based on the above information, it determines whether the task is still under deployment, the validity time, and whether the deployment area and deployment device of the alarm face object code meet the deployment task requirements, thereby determining whether the alarm message is valid. Valid alarm data is written to the large database deployment alarm topic queue for consumption by the view library.

[0106] Figure 6 This is a flowchart of a deployment and management method according to an embodiment of the present invention, such as... Figure 6 As shown, the process includes the following steps:

[0107] Step S301: Obtain the target deployment task and its task information, including the target deployment area and the target deployment object. See step S101 above for details, which will not be elaborated further here.

[0108] Step S302: Based on the target control area, determine the target control system from the connected control systems. See step S102 above for details, which will not be elaborated further here.

[0109] Step S303: If the target control area is larger than a preset range, obtain the large-database control task corresponding to the target control task in the target control system. The large-database control task is used to control multiple control objects within the same target control area. See step S103 above for details, which will not be elaborated further here.

[0110] Step S304: The target object to be controlled is written into the large database control task to control the target object and obtain the control result. See step S104 above for details, which will not be elaborated further here.

[0111] In some optional implementations, the method further includes:

[0112] Step S305: Obtain a preset system assessment task, wherein the system assessment task includes at least one assessment sub-task.

[0113] Step S306: Based on the assessment sub-tasks, assess the connected deployment systems to obtain the assessment results of each connected deployment system corresponding to each assessment sub-task.

[0114] Specifically, step S306 above includes:

[0115] Step S3061: Analyze the assessment sub-tasks to determine the task objectives.

[0116] Specifically, the task objectives include at least one of the following: computational performance assessment, image search performance assessment, recall rate assessment, accuracy rate assessment, diffusion rate assessment, image-based document search compliance rate assessment, trajectory retrieval compliance rate assessment, data reconciliation assessment, document reconciliation consistency rate assessment, and document trajectory reconciliation consistency rate assessment.

[0117] Step S3062: Based on the task objective, obtain the processing result samples and target processing results corresponding to each connected deployment system.

[0118] Step S3063: Based on the target processing results of each connected deployment system and the corresponding processing result samples, obtain the task assessment results of each connected deployment system corresponding to the assessment sub-task.

[0119] Step S307: For each connected deployment system, the task assessment results of all the assessment sub-tasks are merged to obtain the system assessment result of the connected deployment system.

[0120] Step S308: When the system assessment result is that the control system fails to meet the standards, generate corresponding system alarm information.

[0121] The deployment management method provided in this embodiment assesses each connected deployment system through a preset system assessment task. When the system assessment result of a deployment system is that the deployment system fails to meet the standard, a corresponding system alarm message is generated. Therefore, relevant personnel can promptly check the deployment system to ensure the effective execution of deployment of the target task object and improve deployment efficiency and quality.

[0122] It should be noted that with the increasing demand in the domestic and international deployment market and the development of intelligent, integrated, and standardized algorithm models, a single deployment system can no longer fully meet customer needs for analyzing massive video and image resource data. Therefore, it is necessary to introduce the aforementioned multi-engine parsing platform to decouple the application platform from the algorithm services of multiple deployment systems, unify the protocol, and transmit data to the required algorithms according to the protocol between the application and the algorithm, enabling services to process each other through a single standard protocol. Specifically, the multi-algorithm management and scheduling engine system distributes the target deployment tasks and target deployment objects based on the view library to the target deployment area, where the corresponding algorithm library of the deployment system in the target deployment area processes the deployment objects. However, traditional parsing engine applications and algorithms are isolated; if another deployment system's algorithm appears, the former must re-interface with a new algorithm, and may even require reconstruction to ensure normal service, resulting in low efficiency. Therefore, if there are multiple deployment systems, each deployment system needs to be managed and monitored, and the results of management and monitoring should be evaluated regularly. Based on the evaluation of each deployment system, the deployment system used for subsequent task scheduling can be adjusted to achieve optimized scheduling of deployment systems and improve task processing efficiency and accuracy.

[0123] Furthermore, if the task objective is to assess computational performance, step S3062 above includes:

[0124] Step a1: Obtain the computational deployment samples corresponding to each connected deployment system from the assessment sub-tasks.

[0125] Specifically, in practice, a sub-task for performance evaluation can be initiated based on a preset performance evaluation cycle or user operation, with the performance evaluation objective being the performance evaluation sub-task. The performance evaluation control samples include file cover photos and corresponding facial photos. Specifically, performance evaluation control samples can be obtained in the following ways: 20 real-name files with at least 30 newly added tracks in the past three days are randomly selected from control system A, and 10 file cover photos are provided to control systems B and C respectively as performance evaluation control samples. 20 real-name files with at least 30 newly added tracks in the past three days are randomly selected from control system B, and 10 file cover photos are provided to control systems A and C respectively as performance evaluation control samples. Control system C selects 20 real-name files with at least 30 newly added tracks in the past three days from control area a, and provides them to control area b as performance evaluation control samples. The general principle for selecting performance evaluation control samples is that the samples must come from control areas not covered by the corresponding control system to avoid interference. In addition, users can also directly provide performance evaluation control samples.

[0126] Step a2: Each of the computational control samples is sent as a control object to the corresponding control system, so that the corresponding control system can control the computational control samples and obtain the processing result samples corresponding to each connected control system.

[0127] For example, the corresponding control algorithm for generating the archives performs image search on 10 archive cover photos to find 10 images with a similarity of more than 95% to control system A, more than 96% to control system B, and more than 93% to control system C. A total of 100 images are used as the processing result samples. The processing result samples need to be renumbered according to the input control area / control system.

[0128] Step a3: Each of the processing result samples is sent as a control object to the corresponding control system, so that the corresponding control system can control the processing result samples and obtain the target processing result corresponding to each connected control system.

[0129] For example, a control task corresponding to the processing result sample is sent to the corresponding control system, and then the processing result sample is sent as a control object to the corresponding control system. Alarm information generated by the control system is collected to obtain the target processing result corresponding to each connected control system. Then, the face photo corresponding to the processing result sample is sent to the view library.

[0130] Specifically, when the task objective is to assess computational performance, step S3063 includes:

[0131] Based on the preset alarm omission rate assessment rules, calculate the alarm omission rate corresponding to the processing result sample and target processing result of each connected deployment system.

[0132] When the alarm omission rate is less than or equal to the preset alarm omission rate threshold, the corresponding deployment system is determined to have a qualified alarm omission rate for the assessment sub-task.

[0133] Based on the preset alarm accuracy assessment rules, calculate the alarm accuracy of the processing result sample and the target processing result corresponding to each connected deployment system.

[0134] When the alarm accuracy rate is greater than or equal to the preset alarm accuracy rate threshold, the corresponding deployment system is determined to have a qualified alarm accuracy rate for the assessment sub-task.

[0135] Optionally, the preset alarm omission rate threshold is 2%, and the preset alarm accuracy threshold is 96%. In actual operation, an error of 3% is allowed for the alarm omission rate.

[0136] Understandably, performance evaluation includes two aspects: alarm miss rate and alarm accuracy rate.

[0137] The deployment management method provided in this embodiment, when the task objective is computational performance assessment, first obtains computational deployment samples from the assessment sub-tasks, sends these samples to the corresponding deployment systems to obtain processing result samples, and then sends these samples back to the corresponding deployment systems to obtain the target processing result. Therefore, it can effectively utilize the processing result samples and target processing results corresponding to each deployment system to assess the deployment results fed back by the deployment systems and accurately obtain the corresponding computational performance assessment results.

[0138] Furthermore, if the task objective is to evaluate image search performance, step S3062 above further includes:

[0139] Step b1: Call the external image quality detection interface to perform quality detection on the historical images captured in the deployment results fed back by each connected deployment system, and obtain the first quality score of the historical images.

[0140] Step b2: Based on the first quality score of the historical images, select historical images whose first quality score is within a first preset score range from the historical images corresponding to each connected deployment system, and use them as processing result samples corresponding to each connected deployment system.

[0141] It should be noted that the first preset score range needs to be determined based on the actual situation, and is not limited here.

[0142] For example, in practice, image search performance evaluation may include the following four evaluation items: (1 month of hot data) high-quality image retrieval success rate evaluation, (2 months of cold data) high-quality image retrieval success rate evaluation, (1 month of hot data) low-quality image retrieval success rate evaluation, and (2 months of cold data) low-quality image retrieval success rate evaluation. It should be noted that "hot data" and "cold data" refer to different image shooting environments. In practice, processing result samples can be obtained according to a fixed image search performance evaluation cycle. The image search performance evaluation cycle can be set according to needs, such as once a day, starting at 5 AM daily, and executing the evaluation items sequentially. Alternatively, corresponding tasks can be triggered based on user actions. Specifically, the sample of processed results varies depending on the assessment item, as follows: 1. (1-month hot data) The assessment sample for the high-quality image retrieval pass rate assessment is: 50 high-quality images from the past month are randomly selected from the control area managed by the control system, and it is confirmed that there is only one face in the selected high-quality images. The high-quality images refer to images whose image quality score is within the first preset quality range, such as the historical images with the first quality score within the first preset quality range, where the first preset quality range is greater than 80 points. 2. (2-month cold data) The assessment sample for the high-quality image retrieval pass rate assessment is: 25 high-quality images from the past 30 to 60 days and 25 from the past 60 to 90 days are randomly selected from the control area managed by the control system. Drivers and passengers need to select images taken after 9 pm. The selection of images taken after 9 pm is based on the image shooting environment. In actual operation, it can be changed according to actual needs, and is not limited here. 3. (1-month hot data) The assessment sample for the low-quality image retrieval compliance rate is: 50 low-quality images randomly selected from the past month within the control area managed by the system. Drivers and passengers must select images taken after 9 PM. Low-quality images refer to images with a quality score within the second preset quality range, such as historical images with a first quality score within the second preset quality range (50 to 60 points). 4. (2-month cold data) The assessment sample for the low-quality image retrieval compliance rate is: 25 low-quality images randomly selected from the past 30 to 60 days and 25 from 60 to 90 days within the control area managed by the system (excluding cases where images from the responsible control area were not selected). Drivers and passengers must select images taken after 9 PM.

[0143] Step b3: Send each of the processing result samples to the corresponding deployment system for image search to obtain the target processing result corresponding to each connected deployment system.

[0144] Specifically, the obtained processing result samples are used sequentially to call the image search function of the corresponding deployment system, and the time spent on the interface call and whether the original image can be matched in the top five similarity are checked, which is used as the target processing result of the current deployment system.

[0145] Specifically, when the task objective is to evaluate the performance of image search, step S3063 includes:

[0146] Based on the preset image search performance evaluation rules, calculate the image retrieval compliance rate of the processing result samples and target processing results corresponding to each connected deployment system.

[0147] When the image retrieval pass rate is greater than or equal to the preset image retrieval pass rate threshold, the corresponding deployment system is determined to have a qualified image retrieval pass rate for the assessment sub-task.

[0148] Optionally, the preset image retrieval success rate threshold for high-quality images is 95%, and the preset image retrieval success rate threshold for low-quality images is 90%.

[0149] The deployment management method provided in this embodiment, when the task objective is to assess the performance of image search, first extracts the corresponding processing result sample from the deployment results fed back by the deployment system, and then sends the processing result sample to the corresponding deployment system for image search to obtain the target processing result, thus ensuring that the processing result sample can be found in the corresponding deployment system. Therefore, it can effectively utilize the processing result samples and target processing results corresponding to each deployment system to accurately obtain the corresponding image search performance assessment result.

[0150] It should be noted that the performance evaluation of the archive trajectory of the deployment system is mainly divided into the following five evaluation items: recall rate evaluation, accuracy rate evaluation, diffusion rate evaluation, image-based archive search compliance rate evaluation, and archive trajectory retrieval compliance rate evaluation.

[0151] Furthermore, if the task objective is recall rate assessment, step S3062 above further includes:

[0152] Step c1: From the deployment results fed back by each connected deployment system, determine the deployment objects whose trajectories exist within the preset recall rate assessment days for each connected deployment system.

[0153] Optionally, the preset recall rate assessment period is 10 days.

[0154] Specifically, recall rate assessments can be performed according to a preset assessment cycle. The assessment cycle can be set according to needs, such as starting the assessment task every Monday at 8 PM. Then, the Kafka consumer data is stored in a local database, and 20 valid, unregistered files of monitored objects with traces in the past 10 days are randomly selected from the local database according to the monitored area managed by the system.

[0155] Step c2: The monitored objects with existing trajectories are sent to the corresponding monitoring systems for image search to obtain the processing result samples corresponding to each connected monitoring system.

[0156] Specifically, the surveillance system uses the cover photo of the monitored object with the existing trajectory to perform image search, with the similarity threshold set to 90%.

[0157] Step c3: Obtain the trajectory of the monitored object with a trajectory from the corresponding deployment system at the same time and under the same equipment, and use it as the target processing result.

[0158] It should be noted that the target of the recall rate assessment is whether the faces discovered by the control system through image search can be aggregated into files, and the assessment is specifically conducted through sampling.

[0159] Specifically, when the task objective is recall rate assessment, step S3063 includes:

[0160] According to the preset recall rate assessment rules, the processing result samples corresponding to each connected deployment system are compared with the target processing results one by one, and the recall rate is calculated.

[0161] When the recall rate is greater than or equal to the preset recall rate threshold, the corresponding deployment system is determined to have a qualified recall rate for the task assessment result of the assessment sub-task.

[0162] Optionally, the preset recall threshold is 80%.

[0163] It should be noted that comparing the processing result samples corresponding to each connected deployment system with the target processing result is to determine whether the trajectory corresponding to the processing result sample can be found in the target processing result.

[0164] The deployment management method provided in this embodiment, when the task objective is recall rate assessment, first extracts the deployment objects corresponding to the trajectories that meet the recall rate assessment requirements from the deployment results fed back by the deployment system, and sends them to the corresponding deployment system for image search to obtain the corresponding processing result samples. Furthermore, it obtains the trajectory of the deployment object under the same time and device from the corresponding deployment system as the target processing result. Therefore, by utilizing the processing result samples corresponding to each deployment system and the target processing result, it is possible to accurately determine whether the target processing result can be clustered to obtain the corresponding recall rate assessment result.

[0165] Furthermore, if the task objective is accuracy assessment, step S3062 above further includes:

[0166] Step d1: Obtain from the local machine the control objects that have a preset number of trajectories within the preset accuracy assessment days, so as to obtain the processing result samples corresponding to each connected control system.

[0167] Optionally, the preset accuracy assessment period is 10 days, and the preset accuracy assessment quantity is 10 items.

[0168] Specifically, data from Kafka consumer data is stored in a local database, and 20 valid, undeleted files with more than 10 traces in the last 10 days are randomly selected from the local database as samples of the processing results.

[0169] Step d2: Obtain all trajectories of the processing result sample from the corresponding deployment system to obtain the target processing result corresponding to each accessed deployment system.

[0170] Specifically, all trajectories of the processed result samples are obtained from the deployment system, and the image quality of the trajectories is scored. Trajectories with an image quality score higher than 65 are compared with the processed result samples through an external interface, and those with a similarity greater than 90% are considered qualified.

[0171] Specifically, when the task objective is accuracy assessment, step S3063 includes:

[0172] According to the preset accuracy assessment rules, calculate the accuracy of the processing result samples and target processing results corresponding to each connected deployment system.

[0173] When the accuracy rate is greater than or equal to the preset accuracy rate threshold, the corresponding deployment system is determined to have a satisfactory accuracy rate for the assessment result of the assessment sub-task.

[0174] Optionally, the preset accuracy threshold is 95%. In actual operation, an error of 5% is allowed.

[0175] The deployment management method provided in this embodiment, when the task objective is accuracy assessment, first obtains the deployment objects that meet the accuracy assessment requirements for each deployment system from the local machine. These are then used as samples of the processing results. Next, all trajectories corresponding to the processing result samples are obtained from the corresponding deployment systems to obtain the target processing result. Therefore, the processing result samples and the target processing result can be effectively utilized to determine whether the deployment results fed back by the deployment system based on the deployment objects are accurate, thus obtaining the accuracy assessment result.

[0176] Furthermore, if the task objective is to assess the success rate of image-based document search, step S3062 above further includes:

[0177] Step e1: Call the external image quality detection interface to perform quality detection on the archive cover photos of the controlled objects corresponding to each connected local control system, and obtain the second quality score of the archive cover photos.

[0178] Step e2: Based on the second quality score of the file cover photo, select file cover photos whose second quality score is within the second preset score range from the file cover photos corresponding to each connected control system, so as to obtain the processing result sample corresponding to each connected control system.

[0179] Optionally, the second preset score range is greater than 80 points.

[0180] It should be noted that, in actual operation, 100 valid files that have not been cancelled can be randomly selected from the control area under the control system, and the file cover photos with a second quality score of 80 or above can be used as the sample of processing results.

[0181] Step e3: Send each of the processing result samples to the corresponding deployment system for image-based file search to obtain the target processing result corresponding to each connected deployment system.

[0182] Specifically, the processing result sample is used to perform an image search on the corresponding deployment system, and the call time and the first image found in the search match the processing result sample to obtain the target processing result of the corresponding deployment system.

[0183] Specifically, when the task objective is to assess the success rate of image-based document search, step S3063 includes:

[0184] Based on the preset image search compliance rate assessment rules, calculate the image search compliance rate of the processing result sample and the target processing result corresponding to each connected deployment system.

[0185] When the image search compliance rate is greater than or equal to the preset image search compliance rate threshold, the corresponding deployment system is determined to have a qualified image search compliance rate for the assessment sub-task.

[0186] Optionally, the threshold for the image search success rate is 95%. In practice, a 3% error is allowed.

[0187] The deployment management method provided in this embodiment, when the task objective is to assess the compliance rate of image-based document search, first obtains the cover photos of the files of the deployment objects that meet the assessment requirements of image-based document search for each deployment system from the local machine as processing result samples. Then, the processing result samples are sent to the corresponding deployment systems for image-based document search to obtain the corresponding target processing results. Therefore, the image-based document search performance of the deployment system can be effectively evaluated using the processing result samples and the target processing results.

[0188] Furthermore, if the task objective is a diffusion rate assessment, step S3062 above further includes:

[0189] Step f1: Obtain the number of real-name controlled objects corresponding to each connected control system from the local machine, as a sample of the processing results corresponding to each connected control system.

[0190] It should be noted that the real-name monitoring targets here include those whose files have been cancelled, i.e., those whose monitoring has expired.

[0191] Step f2: Obtain the number of identity verification documents corresponding to each connected surveillance system from the local machine, as the target processing result corresponding to each connected surveillance system.

[0192] Specifically, when the task objective is a diffusion rate assessment, step S3063 includes:

[0193] According to the preset diffusion rate assessment rules, the ratio of the processing result sample corresponding to each connected deployment system to the target processing result is calculated to obtain the diffusion rate.

[0194] When the diffusion rate is less than or equal to a preset diffusion rate threshold, the corresponding deployment system is determined to have a diffusion rate qualified for the assessment sub-task.

[0195] Optionally, the preset diffusion rate threshold is 150%. In actual operation, an error of 3% is allowed.

[0196] Understandably, the diffusion rate is calculated by dividing the number of individuals under real-name surveillance by the number of identity verification documents.

[0197] It should be noted that the purpose of the diffusion rate assessment is to determine the extent to which a person under real-name surveillance has multiple files.

[0198] Furthermore, if the task objective is to assess the trajectory retrieval success rate, step S3062 above further includes:

[0199] Step g1: Obtain from the local machine the control objects that have a preset number of trajectory retrieval assessments within the preset trajectory retrieval assessment days, so as to obtain the trajectory retrieval assessment objects corresponding to each connected control system.

[0200] Optionally, the preset trajectory retrieval assessment period is one month, and the preset trajectory retrieval assessment quantity is 5.

[0201] For example, a batch of files of monitored objects are obtained from the local system, grouped into groups for each month of the most recent six months, and monitored objects with more than five trajectories within a month are used as the trajectory retrieval assessment objects of the corresponding monitoring system.

[0202] Step g2: Obtain the trajectory of the trajectory retrieval assessment object from the task issuing object to obtain the processing result sample corresponding to each connected deployment system. The task issuing object is used to issue target deployment tasks.

[0203] For example, five tracks are obtained from the task publishing object (such as a view library).

[0204] Step g3: Obtain the trajectory corresponding to the trajectory retrieval assessment object from the corresponding deployment system to obtain the target processing result corresponding to each accessed deployment system.

[0205] Specifically, when the task objective is to assess the trajectory retrieval success rate, step S3063 includes:

[0206] Check whether the processing result sample corresponding to each accessed deployment system exists in the corresponding target processing result, and check the call time and trajectory hit status of the interface corresponding to the target processing result to obtain the check result. The trajectory hit status is used to indicate the situation where the target processing result hits the corresponding processing result sample.

[0207] Based on the preset trajectory retrieval compliance rate assessment rules and the inspection results, the trajectory retrieval compliance rate corresponding to each connected deployment system is obtained;

[0208] When the trajectory retrieval pass rate is greater than or equal to the preset trajectory pass rate threshold, the corresponding deployment system is determined to have a qualified trajectory retrieval pass rate for the assessment sub-task.

[0209] Optionally, the preset trajectory compliance rate threshold is 99%. In actual operation, a 4% error is allowed in the trajectory retrieval compliance rate.

[0210] It should be noted that the assessment objective of the trajectory retrieval compliance rate is to evaluate the performance of the trajectory retrieval system and whether there are any trajectory loss issues.

[0211] Furthermore, if the task objective is data reconciliation assessment, step S3062 above also includes:

[0212] Step h1: Obtain the number of face images of the monitored area from each connected surveillance system to obtain the processing result samples of each connected surveillance system.

[0213] Step h2: Obtain the number of face images of the monitored areas under the responsibility of each connected monitoring system from the local machine, so as to obtain the target processing results of each connected monitoring system.

[0214] Specifically, when the task objective is data reconciliation assessment, step S3063 includes:

[0215] Based on the preset data reconciliation assessment rules, calculate the data reconciliation consistency rate between the processing result samples and the target processing results corresponding to each connected deployment system;

[0216] When the data reconciliation consistency rate is greater than or equal to the preset data reconciliation consistency rate threshold, the corresponding deployment system is determined to have passed the data reconciliation assessment for the assessment sub-task.

[0217] Optionally, the preset data reconciliation consistency rate threshold is 99.5%. In actual operation, a 1% error in the data reconciliation consistency rate is permissible.

[0218] It should be noted that the assessment objective of the data reconciliation assessment is whether the total number of face images received by each monitoring system each day is consistent with the total number of face images counted by the multi-algorithm engine.

[0219] Furthermore, if the task objective is to assess the consistency rate of file trajectory reconciliation, the above step S3062 further includes:

[0220] Step I1: Obtain from the local machine the control objects that have a preset number of file trajectory consistency rate assessments within the preset number of file trajectory consistency rate assessment days, so as to obtain the consistency rate assessment objects corresponding to each connected control system.

[0221] Optionally, the preset file trajectory consistency rate assessment period is 3 days, and the preset file trajectory consistency rate assessment quantity is 3.

[0222] For example, the multi-control system engine stores Kafka consumer data in a local database. The evaluation sample randomly selects files from the local database containing at least three newly added trajectories within the past three days. The number of files is four times the number of control areas managed by the control system. Finally, the best half of the files are selected to calculate the evaluation score. For instance, if control system A manages six control areas, 24 files of control targets are selected as the input sample, and the 12 best-performing files are chosen to calculate the evaluation score.

[0223] Step I2: Obtain the number of trajectories corresponding to the consistency rate assessment object from the local machine to obtain the processing result samples corresponding to each connected deployment system.

[0224] Step I3: Obtain the number of trajectories corresponding to the consistency rate assessment object from the corresponding deployment system to obtain the target processing result corresponding to each connected deployment system.

[0225] Specifically, when the task objective is to assess the consistency rate of file tracking and reconciliation, the above step S3063 includes:

[0226] Based on the preset file trajectory reconciliation consistency rate assessment rules, calculate the trajectory reconciliation consistency rate between the processing result sample and the target processing result corresponding to each connected deployment system.

[0227] When the trajectory reconciliation consistency rate is greater than or equal to the preset trajectory reconciliation consistency rate threshold, the corresponding deployment system is determined to have a qualified trajectory reconciliation consistency rate for the assessment sub-task.

[0228] Optionally, the preset trajectory reconciliation consistency rate threshold is 99.5%. In actual operation, a 3% error in the consistency rate is permissible.

[0229] It should be noted that the assessment objective of the file trajectory reconciliation consistency rate is whether the number of file trajectories in the control system is consistent with the number of file trajectories in the local storage, and the assessment is conducted by sampling.

[0230] Furthermore, if the task objective is to assess the consistency rate of file reconciliation, step S3062 above also includes:

[0231] Step J1: Obtain the number of valid files for the monitored area under the responsibility of each connected monitoring system, so as to obtain the processing result sample of each connected monitoring system.

[0232] Step J2: Obtain the number of valid files in the control area of ​​each connected control system from the local machine, so as to obtain the target processing result of each connected control system.

[0233] Specifically, when the task objective is to assess the consistency rate of file reconciliation, step S3063 includes:

[0234] Based on the preset file reconciliation consistency rate assessment rules, calculate the file reconciliation consistency rate between the processing result sample and the target processing result corresponding to each connected deployment system.

[0235] When the file reconciliation consistency rate is greater than or equal to the preset file reconciliation consistency rate threshold, the corresponding control system is determined to have passed the file reconciliation consistency rate assessment for the assessment sub-task.

[0236] Optionally, the preset file reconciliation consistency rate threshold is 99.9%. In actual operation, a 5% error in the file reconciliation consistency rate is permissible.

[0237] It should be noted that the assessment target for the file reconciliation consistency rate is whether the number of valid files stored internally in each control system is consistent with the number of valid files stored locally.

[0238] Taking the aforementioned multi-engine parsing platform as an example, the deployment management method of the present invention will be described below with three specific embodiments:

[0239] Example 1, such as Figure 7 As shown, the main tasks of each system in the deployment and control of the monitored objects are as follows:

[0240] View Library:

[0241] 1. Synchronize the deployment information of the four types of deployment systems issued by the user platform, and construct four "deployment information tables" for different deployment systems, allowing the corresponding deployment systems to read, but not write.

[0242] 2. Provide the corresponding real-time facial recognition data to the corresponding surveillance system for analysis.

[0243] 3. Receive alarm information (i.e., control results) generated by the control system after deployment through the corresponding Kafka topic queue of the large database deployment task, and perform subsequent processing.

[0244] 4. It can read the deployment status information table of the corresponding deployment system to confirm the actual deployment status of the deployment object.

[0245] Deployment system:

[0246] 1. Periodically (e.g., every 12 hours) synchronize the data in the "Control Information Table of Controlled Objects" of the preset image library to perform local control. After successful control, write the actual control information into the "Control Status Information Table of Controlled Objects" of this control system to confirm that the corresponding controlled object has been successfully controlled.

[0247] 2. Obtain real-time facial data collected by this surveillance system from the view library and perform surveillance comparison and analysis.

[0248] 3. If the comparison triggers an alarm, the alarm information is sent to the view library through the alarm Kafka topic queue of the corresponding large database control task.

[0249] Multi-algorithm management and scheduling engine system:

[0250] 1. Monitor and reconcile the "Control Information Table of Controlled Objects" synchronized with the view library.

[0251] 2. Monitor and reconcile the control status of the controlled objects with the corresponding control system.

[0252] 3. Monitor and reconcile the generation and receipt of alarm information for monitored targets.

[0253] Example 2, as follows Figure 8 As shown, the main functions of each system in the data aggregation business are as follows:

[0254] View Library:

[0255] 1. Provide a target information table for data aggregation, and update and maintain it according to business needs.

[0256] 2. Provide the corresponding real-time facial recognition data to the corresponding surveillance system for analysis and archiving.

[0257] 3. Receive the archive information and archive trajectory information generated by the deployment system.

[0258] Deployment system:

[0259] 1. Periodically (e.g., every 12 hours) synchronize the data in the aggregated target information table of the view library.

[0260] 2. Obtain real-time facial data collected by this surveillance system from the view library, perform local analysis, file aggregation, and real-name comparison processing.

[0261] 3. Register new files with the view library and send new file trajectory data.

[0262] 4. If there is a file merge operation, then initiate an undo operation for the corresponding file that needs to be undone to the view library.

[0263] Multi-algorithm management and scheduling engine system:

[0264] 1. Monitor and reconcile the data in the target information table of the synchronized archive in the view library.

[0265] 2. Monitor and reconcile the data acquisition status of the target information table with the corresponding deployment system.

[0266] 3. Supervise and reconcile the generation and receipt of archives and trajectory data.

[0267] Example 3, such as Figure 9 As shown, the main functions of each system in the image search service are as follows:

[0268] View Library:

[0269] 1. Send routine business commands such as image search to the multi-algorithm management and scheduling engine system through the interface.

[0270] Deployment system:

[0271] 1. Receive routine business instructions such as image search from the multi-algorithm management and scheduling engine system, and return the execution results to the multi-algorithm management and scheduling engine system.

[0272] Multi-algorithm management and scheduling engine system:

[0273] 1. Receive routine business instructions such as image search from the view library, and forward them to the corresponding deployment system according to the requirements of the instructions.

[0274] 2. After receiving the results returned by the deployment system after executing the command, integrate the results and return them to the view library.

[0275] 3. Supervise and reconcile the process of receiving and sending instructions for the view library and deployment system, as well as the sending and receiving of result data.

[0276] It is worth noting that, firstly, this invention can effectively standardize multiple deployment systems through a multi-algorithm management and scheduling engine system, shielding the differences between them and effectively solving the complexity at the application level. Secondly, this invention can effectively manage the real-time display of the current service status of each deployment system, presenting it to users through a management platform, allowing them to directly view the current service status through page analysis. Thirdly, this invention can effectively assess whether each deployment system meets the standards, issuing timely alarms when the algorithm effect and performance of a deployment system fail to meet the standards, and arranging relevant personnel to investigate and resolve the problem. Fourthly, this invention realizes deployment system task orchestration, determining which deployment system meets the requirements based on protocol specification parsing, and accurately assigning tasks to the compliant deployment system, solving problems such as resource waste and low execution efficiency of each deployment system, and improving the real-time transmission of data streams.

[0277] This embodiment also provides a deployment management device for implementing the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0278] This embodiment provides a deployment management device, such as... Figure 10 As shown, it includes:

[0279] The deployment task acquisition module 401 is used to acquire the target deployment task and its task information, wherein the task information includes the target deployment area and the target deployment object.

[0280] The deployment system selection module 402 is used to determine the target deployment system from the connected deployment systems based on the target deployment area;

[0281] The large warehouse task selection module 403 is used to obtain the large warehouse deployment task corresponding to the target deployment task in the target deployment system if the target deployment area is larger than a preset range. The large warehouse deployment task is used to deploy multiple deployment objects in the same target deployment area.

[0282] The deployment object deployment module 404 is used to write the target deployment object into the large database deployment task in order to deploy the target deployment object and obtain the deployment result.

[0283] In some alternative embodiments, the apparatus further includes:

[0284] The small task selection module is used to send the target deployment task and its target deployment object to the target deployment system if the target deployment area is less than or equal to the preset range, so as to deploy the target deployment object and obtain the deployment result.

[0285] In some alternative embodiments, the apparatus further includes:

[0286] The deployment result receiving module is used to obtain the target task code corresponding to the target deployment result when it receives the target deployment result fed back by any connected deployment system;

[0287] The deployment object determination module is used to obtain the deployment object corresponding to the target deployment result when the target task code is the task code of the large database deployment task;

[0288] The deployment task query module is used to determine the deployment task corresponding to the target deployment result and its task status from the pre-stored deployment tasks corresponding to the large database deployment task, based on the deployment object corresponding to the target deployment result.

[0289] The first result feedback module is used to send the target deployment result to the corresponding business system according to the deployment task corresponding to the target deployment result when the task status meets the preset effective deployment conditions.

[0290] In some alternative embodiments, the apparatus further includes:

[0291] The second result feedback module is used to determine the control task corresponding to the target control result based on the target task code when the target task code is not the task code of the large database control task, so as to send the target control result to the corresponding business system.

[0292] In some optional embodiments, the device further includes: a deployment system assessment module; wherein, the deployment system assessment module includes:

[0293] The assessment task acquisition unit is used to acquire a preset system assessment task, wherein the system assessment task includes at least one assessment sub-task.

[0294] The deployment system assessment unit is used to assess the connected deployment systems based on the assessment sub-tasks, and obtain the assessment results of each connected deployment system corresponding to each assessment sub-task.

[0295] The assessment result fusion unit is used to fuse the task assessment results of all the assessment sub-tasks for each connected deployment system to obtain the system assessment result of the connected deployment system.

[0296] The system assessment alarm unit is used to generate corresponding system alarm information when the system assessment result indicates that the control system fails to meet the standards.

[0297] In some optional implementations, the deployment system assessment unit includes:

[0298] The task parsing subunit is used to parse the assessment subtasks and determine the task objectives;

[0299] The task processing subunit is used to obtain processing result samples and target processing results corresponding to each connected deployment system based on the task objective.

[0300] The assessment subunit is used to obtain the task assessment result of each connected deployment system corresponding to the assessment subtask based on the target processing result of each connected deployment system and the corresponding processing result sample.

[0301] In some optional implementations, the task processing subunit is specifically used for:

[0302] If the task objective is to assess computational performance, computational control samples corresponding to each connected control system are obtained from the assessment sub-tasks; each computational control sample is sent as a control object to the corresponding control system, so that the corresponding control system controls the computational control sample to obtain the processing result sample corresponding to each connected control system; each processing result sample is sent as a control object to the corresponding control system, so that the corresponding control system controls the processing result sample to obtain the target processing result corresponding to each connected control system.

[0303] Alternatively, if the task objective is to assess the performance of image search, an external image quality detection interface is invoked to perform quality detection on the historical images captured in the deployment results fed back by each connected deployment system, thereby obtaining a first quality score for the historical images; based on the first quality score of the historical images, historical images with the first quality score falling within a first preset score range are selected from the historical images corresponding to each connected deployment system, as processing result samples corresponding to each connected deployment system; each processing result sample is then sent to the corresponding deployment system for image search to obtain the target processing result corresponding to each connected deployment system;

[0304] Alternatively, if the task objective is recall rate assessment, the control objects with trajectories within the preset recall rate assessment days are determined from the control results fed back by each connected control system; the control objects with trajectories are sent to the corresponding control system for image search to obtain the processing result samples corresponding to each connected control system; the trajectories of the control objects with trajectories under the same time and the same device are obtained from the corresponding control system as the target processing result;

[0305] Alternatively, if the task objective is accuracy assessment, the system obtains control objects with a preset number of trajectories within a preset number of accuracy assessment days from the local system to obtain processing result samples corresponding to each connected control system; from the corresponding control system, the system obtains all trajectories of the processing result samples to obtain the target processing result corresponding to each connected control system.

[0306] Alternatively, if the task objective is to assess the pass rate of image-based document search, an external image quality detection interface is invoked to perform quality detection on the cover photos of the archives of the monitored objects corresponding to each connected local control system, thereby obtaining a second quality score for the cover photos. Based on the second quality score of the cover photos, cover photos with the second quality score falling within a second preset score range are selected from the cover photos corresponding to each connected control system to obtain processing result samples corresponding to each connected control system. Each processing result sample is then sent to the corresponding control system for image-based document search to obtain the target processing result corresponding to each connected control system.

[0307] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0308] In this embodiment, the deployment management device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0309] This invention also provides a deployment management platform, which has the above-mentioned features. Figure 10 The deployment management device shown is included in the deployment management platform, which comprises:

[0310] A view library is used to issue target deployment tasks and their task information, the task information including the target deployment area and the target deployment object;

[0311] A multi-algorithm management and scheduling engine system, which is connected to the view library, is used to execute the deployment management method of any of the above embodiments;

[0312] At least one deployment system is connected to the multi-algorithm management and scheduling engine system, and is used to deploy the target deployment object under the scheduling of the multi-algorithm management and scheduling engine system to obtain deployment results.

[0313] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0314] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and all such modifications and variations fall within the scope defined by the appended claims.

Claims

1. A deployment and management method, characterized in that, The method includes: Acquire the target deployment task and its task information, wherein the task information includes the target deployment area and the target deployment object; Based on the target control area, the target control system is determined from the connected control systems; If the target control area is larger than a preset range, obtain the large-database control task corresponding to the target control task in the target control system. The large-database control task is used to control multiple control objects within the same target control area. The large-database control task has the following attribute requirements: the large-database control task is always valid and only needs to be created once; the task code is fixed; the control area is larger than the preset range; the control threshold is fixed; the control time is always valid; the task control status is always "controlling"; and the task code of the large-database control task needs to be disclosed to the task issuing object. The target control object is written into the large database control task to control the target control object and obtain the control result; wherein, the target control task is converted into the control object of the corresponding large database control task, the target control object is assigned the task code of the corresponding large database control task, and the target control task is sent to the control system in the manner of the large database control task. If the target control area is less than or equal to the preset range, the target control task and its target control object are sent to the target control system to control the target control object and obtain the control result; wherein, the target control object is sent to the corresponding target control area to call the corresponding control system for processing.

2. The method according to claim 1, characterized in that, The method further includes: When a target deployment result is received from any of the connected deployment systems, the target task code corresponding to the target deployment result is obtained; When the target task is encoded as the task code of the large database deployment task, the deployment object corresponding to the target deployment result is obtained; Based on the control object corresponding to the target control result, determine the control task and its status corresponding to the target control result from the pre-stored control tasks corresponding to the large warehouse control task; When the task status meets the preset effective deployment conditions, the target deployment result is sent to the corresponding business system according to the deployment task corresponding to the target deployment result.

3. The method according to claim 2, characterized in that, The method further includes: When the target task code is not the task code of the large database deployment task, the deployment task corresponding to the target deployment result is determined according to the target task code, so as to send the target deployment result to the corresponding business system.

4. The method according to claim 1, characterized in that, The method further includes: Obtain a preset system assessment task, wherein the system assessment task includes at least one assessment sub-task; The accessed deployment system is evaluated based on the evaluation sub-tasks to obtain the task evaluation results of each accessed deployment system corresponding to each evaluation sub-task. For each connected deployment system, the task assessment results of all the assessment sub-tasks are merged to obtain the system assessment result of the connected deployment system. When the system assessment result indicates that the control system fails to meet the standards, a corresponding system alarm message is generated.

5. The method according to claim 4, characterized in that, The assessment of the connected deployment systems based on the assessment sub-tasks, to obtain the task assessment results for each connected deployment system corresponding to each assessment sub-task, includes: The assessment sub-tasks are analyzed to determine the task objectives; Based on the task objective, obtain the processing result samples and target processing results corresponding to each connected deployment system; Based on the target processing results of each connected deployment system and the corresponding processing result samples, the task assessment results of each connected deployment system corresponding to the assessment sub-task are obtained.

6. The method according to claim 5, characterized in that, The step of obtaining processing result samples and target processing results corresponding to each connected deployment system based on the task objective includes: If the task objective is to assess computational performance, computational control samples corresponding to each connected control system are obtained from the assessment sub-tasks; each computational control sample is sent as a control object to the corresponding control system, so that the corresponding control system controls the computational control sample to obtain the processing result sample corresponding to each connected control system; each processing result sample is sent as a control object to the corresponding control system, so that the corresponding control system controls the processing result sample to obtain the target processing result corresponding to each connected control system. Alternatively, if the task objective is to assess the performance of image search, an external image quality detection interface is invoked to perform quality detection on the historical images captured in the deployment results fed back by each connected deployment system, thereby obtaining a first quality score for the historical images; based on the first quality score of the historical images, historical images with the first quality score falling within a first preset score range are selected from the historical images corresponding to each connected deployment system, as processing result samples corresponding to each connected deployment system; each processing result sample is then sent to the corresponding deployment system for image search to obtain the target processing result corresponding to each connected deployment system; Alternatively, if the task objective is recall rate assessment, the control objects with trajectories within a preset recall rate assessment period are determined from the control results fed back by each connected control system; the control objects with trajectories are sent to the corresponding control system for image search to obtain the processing result samples corresponding to each connected control system; the trajectories of the control objects with trajectories under the same time and the same device are obtained from the corresponding control system as the target processing result; Alternatively, if the task objective is accuracy assessment, the system obtains control objects with a preset number of trajectories within a preset number of accuracy assessment days from the local system to obtain processing result samples corresponding to each connected control system; from the corresponding control system, the system obtains all trajectories of the processing result samples to obtain the target processing result corresponding to each connected control system. Alternatively, if the task objective is to assess the pass rate of image-based document search, an external image quality detection interface is invoked to perform quality detection on the cover photos of the archives of the monitored objects corresponding to each connected local control system, thereby obtaining a second quality score for the cover photos. Based on the second quality score of the cover photos, cover photos with the second quality score falling within a second preset score range are selected from the cover photos corresponding to each connected control system to obtain processing result samples corresponding to each connected control system. Each processing result sample is then sent to the corresponding control system for image-based document search to obtain the target processing result corresponding to each connected control system.

7. A deployment and management device, characterized in that, The device includes: The deployment task acquisition module is used to acquire target deployment tasks and their task information, wherein the task information includes target deployment area and target deployment object; The deployment system selection module is used to determine the target deployment system from the connected deployment systems based on the target deployment area; The large-database task selection module is used to obtain the large-database deployment task corresponding to the target deployment task in the target deployment system if the target deployment area is larger than a preset range. The large-database deployment task is used to deploy multiple deployment objects in the same target deployment area. The large-database deployment task has the following attribute requirements: the large-database deployment task is always valid and only needs to be created once, the task code is fixed, the deployment area is larger than the preset range, the deployment threshold is fixed, the deployment time is always valid, the task deployment status is always in deployment, and the task code of the large-database deployment task needs to be informed to the task issuing object. The deployment object deployment module is used to write the target deployment object into the large database deployment task to deploy the target deployment object and obtain the deployment result; wherein, the target deployment task is converted into the deployment object of the corresponding large database deployment task, the target deployment object is assigned the task code of the corresponding large database deployment task, and the target deployment task is sent to the deployment system in the manner of the large database deployment task; The small task selection module is used to send the target deployment task and its target deployment object to the target deployment system if the target deployment area is less than or equal to the preset range, so as to deploy the target deployment object and obtain the deployment result; wherein, the target deployment object is sent to the corresponding target deployment area to call the corresponding deployment system for processing.

8. A deployment management platform, characterized in that, include: A view library is used to issue target deployment tasks and their task information, the task information including the target deployment area and the target deployment object; A multi-algorithm management and scheduling engine system, which is connected to the view library, is used to execute the deployment management method according to any one of claims 1 to 6; At least one deployment system is connected to the multi-algorithm management and scheduling engine system, and is used to deploy the target deployment object under the scheduling of the multi-algorithm management and scheduling engine system to obtain deployment results.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the deployment management method according to any one of claims 1 to 6.

Images