Multi-source remote sensing image intelligent interpretation scheduling system and method

Abstract

The embodiment of the application provides a kind of multi-source remote sensing image intelligent interpretation scheduling system and method, can be applied to remote sensing image processing technical field, the system includes: algorithm management module, for the standardization encapsulation of multiple independent multi-source remote sensing information intelligent interpretation algorithm respectively, deployment is as independent algorithm service, form algorithm resource pool;Algorithm service scheduling module is connected with algorithm management module, for calling at least one algorithm service from algorithm resource pool, and according to the scheduling strategy of algorithm service preset algorithm service is scheduled to the computing node in the preset computing cluster and runs, to complete interpretation task, generates interpretation result;Algorithm service evaluation and feedback module is connected with algorithm service scheduling module, for according to interpretation result, the processing effect of algorithm service is evaluated, and evaluation result is generated;Algorithm service scheduling module is also used to dynamically update the scheduling strategy of algorithm service according to evaluation result.

Description

Technical Field

[0001] This invention relates to the field of remote sensing image processing technology, and more specifically to a multi-source remote sensing image intelligent interpretation and scheduling system and method. Background Technology

[0002] In recent years, with the rapid development of remote sensing satellite research and application in my country, multiple satellite constellations led by commercial aerospace companies have been successively deployed in orbit, with planned numbers ranging from dozens to hundreds of satellites. As the amount of data from in-orbit satellites increases, the fusion and application of massive satellite data has become an urgent problem to be solved. In particular, for intelligent interpretation data generated by different remote sensing satellite payloads, the fusion of multi-source remote sensing image processing faces challenges such as inconsistent formats and data heterogeneity due to differences in processing algorithms, data types, and evaluation methods. Existing fusion engines mainly integrate single algorithm types and process and apply the algorithm interpretation results from single data sources, lacking evaluation and feedback of the algorithms, making it difficult to meet the demands of multi-sensor, massive algorithm processing. Summary of the Invention

[0003] In view of the above problems, the present invention provides a multi-source remote sensing image intelligent interpretation and scheduling system and method.

[0004] According to a first aspect of the present invention, a multi-source remote sensing image intelligent interpretation and scheduling system is provided, comprising:

[0005] The algorithm management module is used to standardize and encapsulate multiple independent multi-source remote sensing information intelligent interpretation algorithms, deploy them as independent algorithm services, and form an algorithm resource pool, wherein the algorithm resource pool includes the independent algorithm services;

[0006] The algorithm service scheduling module is communicatively connected to the algorithm management module. It is used to call at least one algorithm service from the algorithm resource pool and schedule the algorithm service to a computing node in a preset computing cluster according to the algorithm service's preset scheduling strategy, so as to complete the interpretation task and generate interpretation results.

[0007] An algorithm service evaluation and feedback module, connected to the algorithm service scheduling module, is used to evaluate the processing effect of the algorithm service based on the interpretation results and generate evaluation results.

[0008] The algorithm service scheduling module is also used to dynamically update the scheduling strategy of the algorithm service based on the evaluation results.

[0009] According to an embodiment of the present invention, it further includes:

[0010] The common service module is communicatively connected to the algorithm service scheduling module and the algorithm service evaluation and feedback module, and is used to uniformly store and manage the interpretation results and the basic data required to complete the interpretation task;

[0011] The algorithm service evaluation and feedback module obtains the interpretation results through the common service module.

[0012] According to an embodiment of the present invention, the step of standardizing and encapsulating multiple independent multi-source remote sensing information intelligent interpretation algorithms and deploying them as independent algorithm services includes:

[0013] The interpretation algorithm is encapsulated into an independent service image using Docker container technology;

[0014] Based on the independent service image, a corresponding algorithm service template is generated;

[0015] Based on the processing requirements of the interpretation task, the algorithm service template is deployed to instantiate the algorithm service.

[0016] According to an embodiment of the present invention, the algorithm management module is further configured to:

[0017] Provide an overview of storage resources, including at least one of the following: algorithm service version, memory usage, storage space capacity, total number of directories, total number of algorithms, available capacity, and total capacity;

[0018] Provide algorithm status statistics, including at least one of the following: the number of times the algorithm service is called and the number of tasks.

[0019] According to an embodiment of the present invention, the algorithm management module is further configured to:

[0020] The algorithm service template is managed, and the management operation includes at least one of viewing, editing and deleting the algorithm service template;

[0021] The algorithm service template contains configuration information for deployment, including at least one of the following: algorithm name, version, status, image size, and addition time.

[0022] According to an embodiment of the present invention, scheduling the algorithm service to a computing node in a preset computing cluster for operation according to a pre-defined scheduling strategy includes:

[0023] Based on the CPU, GPU, and memory resource requirements and business priorities of the algorithm service, a target computing node is determined from the list of computing nodes for the algorithm service container to be scheduled.

[0024] The binding relationship information between the algorithm service container and the target computing node is written into a distributed key-value storage system;

[0025] The binding relationship information is obtained by using the proxy service deployed on the target computing node to monitor the distributed key-value storage system;

[0026] The proxy service is used to pull the corresponding algorithm service image based on the binding relationship information and instantiate it to start the algorithm service container.

[0027] According to an embodiment of the present invention, the algorithm management module is further configured to:

[0028] Based on the RocketMQ message queue mechanism, asynchronous information interaction between the independent algorithm services is achieved.

[0029] According to an embodiment of the present invention, the step of dynamically updating the scheduling strategy of the algorithm service based on the evaluation result includes:

[0030] Based on the historical evaluation results generated by the algorithm service evaluation and feedback module, algorithm services with a comprehensive score higher than a preset threshold are prioritized for use; wherein, the historical evaluation results are generated based on the quantification of multiple performance indicators.

[0031] A second aspect of the present invention provides an intelligent interpretation and scheduling method for multi-source remote sensing images, the method comprising:

[0032] Multiple independent multi-source remote sensing information intelligent interpretation algorithms are standardized and packaged, and deployed as independent algorithm services to form an algorithm resource pool containing the independent algorithm services.

[0033] At least one algorithm service is invoked from the algorithm resource pool, and according to the algorithm service's preset scheduling strategy, the algorithm service is scheduled to run on a computing node in a preset computing cluster to complete the interpretation task and generate interpretation results.

[0034] Based on the interpretation results, the processing effect of the algorithm service is evaluated, and an evaluation result is generated;

[0035] Based on the evaluation results, the scheduling strategy of the algorithm service is dynamically updated.

[0036] According to an embodiment of the present invention, it further includes:

[0037] The interpretation results and the basic data required to complete the interpretation task are stored and managed in a unified manner.

[0038] The intelligent interpretation and scheduling system and method for multi-source remote sensing images provided in this invention can achieve at least some of the following technical effects:

[0039] (1) Through standardized design, the integration and scheduling of different remote sensing fusion algorithms are realized, thereby realizing the fusion of multi-source remote sensing information;

[0040] (2) An algorithm evaluation feedback mechanism has been added. Based on the algorithm evaluation results, the algorithm scheduling strategy is automatically adjusted to improve the accuracy of multi-source remote sensing information processing. At the same time, the algorithm can be iteratively updated according to the evaluation results. Attached Figure Description

[0041] The above-described features, other objects, and advantages of the present invention will become clearer from the following description of embodiments of the invention with reference to the accompanying drawings, in which:

[0042] Figure 1 This schematic diagram illustrates the structure of a multi-source remote sensing image intelligent interpretation and scheduling system according to an embodiment of the present invention.

[0043] Figure 2 A flowchart illustrating the algorithmic service evaluation process according to an embodiment of the present invention is shown schematically.

[0044] Figure 3 A flowchart illustrating a multi-source remote sensing image intelligent interpretation and scheduling method according to an embodiment of the present invention is shown. Detailed Implementation

[0045] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the invention. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the invention for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0046] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0047] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0048] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0049] Figure 1 The diagram illustrates the structure of a multi-source remote sensing image intelligent interpretation and scheduling system according to an embodiment of the present invention. Figure 2 A flowchart illustrating the algorithmic service evaluation process according to an embodiment of the present invention is shown.

[0050] like Figure 1 and 2 As shown, the multi-source remote sensing image intelligent interpretation scheduling system of this embodiment includes an algorithm management module 310, an algorithm service scheduling module 320, and an algorithm service evaluation and feedback module 330. The algorithm management module 310 is used to standardize and encapsulate multiple independent multi-source remote sensing information intelligent interpretation algorithms, deploying them as independent algorithm services to form an algorithm resource pool, which includes independent algorithm services. The algorithm service scheduling module 320 is communicatively connected to the algorithm management module 310, and is used to call at least one algorithm service from the algorithm resource pool, and schedule the algorithm service to a computing node in a preset computing cluster according to the algorithm service's pre-set scheduling strategy to complete the interpretation task and generate interpretation results. The algorithm service evaluation and feedback module 330, together with the algorithm service scheduling module 320, is used to evaluate the processing effect of the algorithm service based on the interpretation results and generate evaluation results. The algorithm service scheduling module 320 is also used to dynamically update the algorithm service scheduling strategy based on the evaluation results.

[0051] Standardized packaging refers to packaging an algorithm and its dependent runtime environment into a single, portable image file using Docker container technology. This process may include: creating a Dockerfile for each algorithm, defining the base image, copying the algorithm code and model, installing dependent libraries, and setting up startup commands; building the algorithm service image using the Docker engine, and uploading it to a private image repository for unified version management. After packaging in this way, each algorithm becomes an isolated, independent service that can be started and stopped at any time.

[0052] The pre-defined scheduling strategy may initially include a strategy that comprehensively calculates based on the resource requirements of the algorithm service (such as the number of CPU cores, GPUs, and memory size) and business priorities (such as task urgency and algorithm version stability). The algorithm service scheduling module 320 selects computing nodes that meet the resource requirements and have suitable loads for the service containers to be run according to this strategy.

[0053] Evaluation based on the interpretation results can refer to the algorithm service evaluation and feedback module 330 receiving the interpretation results (such as target recognition boxes, land cover classification maps, etc.), comparing them with the ground truth data generated by manual interpretation, and calculating one or more quantitative evaluation indicators, such as accuracy, recall, and intersection-union ratio, thereby generating a score for the algorithm execution.

[0054] The scheduling strategy for dynamically updated algorithm services can be achieved by the algorithm service scheduling module 320 maintaining an algorithm rating record table that records the average rating of each algorithm's historical calls. When a new scheduling decision is needed, the scheduling module will refer to this rating record and assign higher weights or priorities to algorithm services with higher historical ratings, making them the first choice during resource contention, thereby achieving automated iterative optimization through "survival of the fittest".

[0055] The invention will now be described in further detail with reference to a specific application scenario.

[0056] Suppose the system needs to integrate two functionally similar algorithms, "Ship Detection Algorithm A" and "Ship Detection Algorithm B".

[0057] First, the algorithm management module 310 uses Docker containers to standardize and encapsulate algorithm A and algorithm B respectively, generating two independent algorithm service images, and registers them into the system's algorithm resource pool.

[0058] When a user submits a remote sensing image of a port area for a ship detection task, the algorithm service scheduling module 320 calls an algorithm service from the resource pool. Initially, the scheduling strategy may be set to round-robin or based on a default priority. Assume that algorithm A is scheduled this time. After algorithm A runs on the computing node, the generated ship position information (interpretation results) is sent to the algorithm service evaluation and feedback module 330.

[0059] Subsequently, the module compares the results of Algorithm A with the manually labeled ground truth data, calculates its accuracy rate to be 85%, and generates the evaluation result.

[0060] Next, the scheduling module receives this evaluation result and updates its internal policy table. Assume that the historical average score of Algorithm A is updated to 85 points. Subsequently, when a similar ship detection task is received again, the dynamic update mechanism of the algorithm service scheduling module 320 comes into play. If the historical score of Algorithm B (e.g., 70 points) is lower than that of Algorithm A, the system will prioritize scheduling the higher-performing Algorithm A to execute the task.

[0061] Through the above process, the system achieves automated algorithm selection and scheduling optimization based on actual results, forming an efficient closed-loop feedback system that significantly improves the accuracy and reliability of intelligent interpretation of multi-source remote sensing images.

[0062] In some embodiments, the multi-source remote sensing image intelligent interpretation and scheduling system further includes a common service module, which is communicatively connected to the algorithm service scheduling module 320 and the algorithm service evaluation and feedback module 330. The common service module is used for unified storage and management of interpretation results and basic data required to complete the interpretation task.

[0063] The algorithm service evaluation and feedback module 330 obtains the interpretation results through the common service module. The interpretation results may include the latitude and longitude information of the detected and identified target location, time, etc., the basic data may include basic geographic information, and the target information may include target category information, etc.

[0064] Building upon the aforementioned ship inspection implementation example, the introduction of a common service module will make the system's data flow clearer and more standardized:

[0065] After the algorithm service scheduling module 320 schedules algorithm A to complete the interpretation task, it will not directly pass the results to the evaluation module. Instead, it will send and store the ship detection results (including target box coordinates, confidence level, etc.) output by algorithm A together with the task metadata in the designated database of the common service module.

[0066] Subsequently, when the algorithm service evaluation and feedback module 330 needs to evaluate the result of the task, it will proactively send a query request to the common service module and obtain the complete interpretation result generated by algorithm A by calling its RESTful API.

[0067] Meanwhile, the evaluation module may also obtain the original remote sensing imagery corresponding to the task from the common service module as a reference, and combine it with the basic vector map of the port area queried from this module (as an auxiliary ground truth) to conduct a comprehensive analysis and evaluation.

[0068] According to embodiments of the present invention, by introducing a common service module, the system can achieve centralized management and service provision of data, enabling core components such as the algorithm service scheduling module 320 and the algorithm service evaluation and feedback module 330 to focus on their own business processes. This achieves decoupling of data and business, enhancing system stability and the independence between modules. This further confirms the advanced nature and rationality of the system architecture of the present invention.

[0069] In some embodiments, the algorithm management module 310 can implement various management and control functions for algorithms based on algorithm standardization, including at least some of algorithm standardization, user and permission management, storage management, and template management.

[0070] The algorithm standardization function can be used for intelligent interpretation algorithms of multi-source remote sensing information of various types and granularities. By standardizing and encapsulating these interpretation algorithms into services, and then collaboratively implementing algorithm application services, it meets the requirements for algorithm reusability and customizability. For algorithm encapsulation, a Docker-based operating system virtualization service is used to instantiate services, build customized algorithm service templates, and upload them to an image repository for unified management. Based on processing needs, the algorithm service templates are used for rapid deployment. For algorithm communication, a RocketMQ message queue mechanism can be used to achieve information exchange between algorithm services through distributed message transmission. Communication between algorithms is asynchronous, reducing interaction coupling.

[0071] User and permission management features provide a unified user identity, enabling fine-grained access control, security auditing, and support for user information storage, CRUD operations, role management, and user-role relationship mapping. User management offers basic operations such as creating, deleting, modifying, and querying user accounts, maintaining basic user account information such as username, password, role, and permissions. Role management manages information for roles such as administrators, general accounts, and operations personnel, allowing users to create custom roles and modify their permission associations, such as user-role relationships, role-service relationships, role CRUD operations, role identifiers, and role permissions. Security auditing records user and service behavior, categorizes and labels it, and provides user operation history records, queries, and auditing capabilities, facilitating troubleshooting of problematic accounts when issues arise.

[0072] The storage management function provides an overview of storage resources and algorithm status statistics. The overview includes algorithm service version, memory usage, storage space capacity, total number of directories, total number of algorithms, available capacity, and total capacity. The status statistics allow users to view the status of various algorithms, including the number of calls and tasks.

[0073] The template management function provides basic capabilities such as viewing, editing, and deleting algorithm service templates. Algorithm service templates enable rapid deployment of algorithm services. Template management information includes algorithm name, version, status, image size, and addition time. After the algorithm service is deployed to the engine, it can be directly invoked from the engine system.

[0074] In some embodiments, the algorithm service scheduling module 320 can bind containers to be scheduled to a suitable compute node in the cluster according to a specific scheduling algorithm and strategy, and write the binding information into a key-value store. Through the algorithm scheduling engine, a suitable compute node is selected from the compute node list for each container in the list of containers to be scheduled, based on CPU / GPU / memory requirements and algorithm priority. Subsequently, the corresponding service on the target compute node listens for the container binding event generated by resource scheduling through a unified interface, then obtains the corresponding container, downloads the container image, and starts the container.

[0075] In some embodiments, based on the historical evaluation results generated by the algorithm service evaluation and feedback module 330, algorithm services with a comprehensive score higher than a preset threshold can be preferentially called. The historical evaluation results are generated based on the quantification of multiple performance indicators.

[0076] Figure 3 A flowchart illustrating a multi-source remote sensing image intelligent interpretation and scheduling method according to an embodiment of the present invention is shown.

[0077] like Figure 3 As shown, the method includes operations S210 to S240.

[0078] In operating S210, multiple independent multi-source remote sensing information intelligent interpretation algorithms are standardized and packaged, and deployed as independent algorithm services to form an algorithm resource pool containing independent algorithm services.

[0079] In operation S220, at least one algorithm service is invoked from the algorithm resource pool, and according to the algorithm service's preset scheduling strategy, the algorithm service is scheduled to run on a computing node in a preset computing cluster to complete the interpretation task and generate interpretation results.

[0080] In operation S230, the processing effect of the algorithm service is evaluated based on the interpretation results, and an evaluation result is generated.

[0081] In operation S240, the scheduling strategy of the algorithm service is dynamically updated based on the evaluation results.

[0082] According to an embodiment of the present invention, the method further includes: uniformly storing and managing the interpretation results and the basic data required to complete the interpretation task.

[0083] According to an embodiment of the present invention, the standardization and encapsulation of multiple independent multi-source remote sensing information intelligent interpretation algorithms and their deployment as independent algorithm services include: encapsulating the interpretation algorithms into independent service images based on Docker container technology; generating corresponding algorithm service templates based on the independent service images; and deploying the algorithm service templates according to the processing requirements of the interpretation task to instantiate them as algorithm services.

[0084] According to an embodiment of the present invention, the method further includes: providing storage resource overview information, including at least one of algorithm service version, memory usage, storage space capacity, total number of directories, total number of algorithms, available capacity and total capacity; and providing algorithm status statistics, including at least one of algorithm service call count and task count.

[0085] According to an embodiment of the present invention, the method further includes: managing algorithm service templates, wherein the management operation includes at least one of viewing, editing and deleting algorithm service templates; wherein the algorithm service template contains configuration information for deployment, and the configuration information includes at least one of algorithm name, version, status, image size and addition time.

[0086] According to an embodiment of the present invention, scheduling an algorithm service to a computing node in a preset computing cluster for execution according to a pre-defined scheduling strategy includes: determining a target computing node from a list of computing nodes for the algorithm service container to be scheduled based on the pre-defined CPU, GPU, and memory resource requirements and business priorities of the algorithm service; writing the binding relationship information between the algorithm service container and the target computing node into a distributed key-value storage system; obtaining the binding relationship information by monitoring the distributed key-value storage system using a proxy service deployed on the target computing node; and using the proxy service to pull the corresponding algorithm service image and instantiate it to start the algorithm service container based on the binding relationship information.

[0087] According to an embodiment of the present invention, the method further includes: implementing asynchronous information interaction between independent algorithm services based on the RocketMQ message queue mechanism.

[0088] According to an embodiment of the present invention, the scheduling strategy for dynamically updating algorithm services based on evaluation results includes: prioritizing the use of algorithm services with a comprehensive score higher than a preset threshold based on historical evaluation results generated by the algorithm service evaluation and feedback module 330; wherein, the historical evaluation results are generated based on the quantification of multiple performance indicators.

[0089] According to embodiments of the present invention, any multiple modules among the algorithm management module 310, algorithm service scheduling module 320, and algorithm service evaluation and feedback module 330 can be merged into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module.

[0090] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0091] Those skilled in the art will understand that the features described in the various embodiments of the present invention can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, the features described in the various embodiments of the present invention can be combined and / or combined in various ways without departing from the spirit and teachings of the present invention. All such combinations and / or combinations fall within the scope of the present invention.

Claims

1. A multi-source remote sensing image intelligent interpretation and scheduling system, characterized in that, include: The algorithm management module is used to standardize and encapsulate multiple independent multi-source remote sensing information intelligent interpretation algorithms, deploy them as independent algorithm services, and form an algorithm resource pool, wherein the algorithm resource pool includes the independent algorithm services; The algorithm service scheduling module is communicatively connected to the algorithm management module. It is used to call at least one algorithm service from the algorithm resource pool and schedule the algorithm service to a computing node in a preset computing cluster according to the algorithm service's preset scheduling strategy, so as to complete the interpretation task and generate interpretation results. An algorithm service evaluation and feedback module, connected to the algorithm service scheduling module, is used to evaluate the processing effect of the algorithm service based on the interpretation results and generate evaluation results. The algorithm service scheduling module is also used to dynamically update the scheduling strategy of the algorithm service based on the evaluation results.

2. The intelligent interpretation and scheduling system for multi-source remote sensing images according to claim 1, characterized in that, Also includes: The common service module is communicatively connected to the algorithm service scheduling module and the algorithm service evaluation and feedback module, and is used to uniformly store and manage the interpretation results and the basic data required to complete the interpretation task; The algorithm service evaluation and feedback module obtains the interpretation results through the common service module.

3. The intelligent interpretation and scheduling system for multi-source remote sensing images according to claim 1, characterized in that, The step of standardizing and encapsulating multiple independent multi-source remote sensing information intelligent interpretation algorithms and deploying them as independent algorithm services includes: The interpretation algorithm is encapsulated into an independent service image using Docker container technology; Based on the independent service image, a corresponding algorithm service template is generated; Based on the processing requirements of the interpretation task, the algorithm service template is deployed to instantiate the algorithm service.

4. The intelligent interpretation and scheduling system for multi-source remote sensing images according to claim 1, characterized in that, The algorithm management module is also used for: Provide an overview of storage resources, including at least one of the following: algorithm service version, memory usage, storage space capacity, total number of directories, total number of algorithms, available capacity, and total capacity; Provide algorithm status statistics, including at least one of the following: the number of times the algorithm service is called and the number of tasks.

5. The intelligent interpretation and scheduling system for multi-source remote sensing images according to claim 3, characterized in that, The algorithm management module is also used for: The algorithm service template is managed, and the management operation includes at least one of viewing, editing and deleting the algorithm service template; The algorithm service template contains configuration information for deployment, including at least one of the following: algorithm name, version, status, image size, and addition time.

6. The intelligent interpretation and scheduling system for multi-source remote sensing images according to claim 1, characterized in that, The step of scheduling the algorithm service to a computing node in a preset computing cluster according to the preset scheduling strategy of the algorithm service includes: Based on the CPU, GPU, and memory resource requirements and business priorities of the algorithm service, a target computing node is determined from the list of computing nodes for the algorithm service container to be scheduled. The binding relationship information between the algorithm service container and the target computing node is written into a distributed key-value storage system; The binding relationship information is obtained by using the proxy service deployed on the target computing node to monitor the distributed key-value storage system; The proxy service is used to pull the corresponding algorithm service image based on the binding relationship information and instantiate it to start the algorithm service container.

7. The intelligent interpretation and scheduling system for multi-source remote sensing images according to claim 1, characterized in that, The algorithm management module is also used for: Based on the RocketMQ message queue mechanism, asynchronous information interaction between the independent algorithm services is achieved.

8. The intelligent interpretation and scheduling system for multi-source remote sensing images according to claim 1, characterized in that, The scheduling strategy for dynamically updating the algorithm service based on the evaluation results includes: Based on the historical evaluation results generated by the algorithm service evaluation and feedback module, algorithm services with a comprehensive score higher than a preset threshold are prioritized for use; wherein, the historical evaluation results are generated based on the quantification of multiple performance indicators.

9. A method for intelligent interpretation and scheduling of multi-source remote sensing images, characterized in that, The method includes: Multiple independent multi-source remote sensing information intelligent interpretation algorithms are standardized and packaged, and deployed as independent algorithm services to form an algorithm resource pool containing the independent algorithm services. At least one algorithm service is invoked from the algorithm resource pool, and according to the algorithm service's preset scheduling strategy, the algorithm service is scheduled to run on a computing node in a preset computing cluster to complete the interpretation task and generate interpretation results. Based on the interpretation results, the processing effect of the algorithm service is evaluated, and an evaluation result is generated; Based on the evaluation results, the scheduling strategy of the algorithm service is dynamically updated.

10. The intelligent interpretation and scheduling method for multi-source remote sensing images according to claim 9, characterized in that, Also includes: The interpretation results and the basic data required to complete the interpretation task are stored and managed in a unified manner.

Images