CN-121999092-A - Data processing method, apparatus, computer device and readable storage medium

CN121999092ACN 121999092 ACN121999092 ACN 121999092ACN-121999092-A

Abstract

The application relates to a data processing method, a data processing device, a computer device and a readable storage medium. The method comprises the steps of obtaining an initial scene image generated by a geographic information system in response to a target operation instruction, controlling a visual verification big model based on a master control model to conduct multi-dimensional image verification of the initial scene image layer by layer to obtain a structured verification result of the initial scene image, determining target deviation rectifying logic corresponding to abnormal information in verification dimensions of images of all layers according to the master control model, the abnormal information and a preset deviation rectifying strategy if the structured verification result is abnormal information, controlling the geographic information system to generate an updated scene image according to the master control model and the target deviation rectifying logic until the updated scene image meets preset iteration conditions to obtain a target scene image, and conducting data processing based on the target scene image to obtain a data processing result. By adopting the method, the accuracy of data processing of the geographic intelligent agent can be improved.

Inventors

WANG KAISONG
LIU YANSONG
LIU GUANG
LUAN WENBO
Ke Jichang
LI LEI
LIU PENG
YI ZHILI
YAN NING
ZENG YANYAN

Assignees

北京市测绘设计研究院

Dates

Publication Date: 20260508
Application Date: 20260410

Claims (10)

1. A data processing method, wherein the method is applied to a geographic agent, the geographic agent comprising a master control model and a visual verification model, the method comprising: acquiring an initial scene image generated by a geographic information system in response to a target operation instruction; Controlling the visual verification model to perform hierarchical multi-dimensional image verification on the initial scene image based on the master control model to obtain a structured verification result of the initial scene image; If the structured verification result is that abnormal information exists, determining target deviation rectifying logic corresponding to the abnormal information in each level of image verification dimension according to the main control model, the abnormal information and a preset deviation rectifying strategy; According to the master control model and the target deviation rectifying logic, controlling the geographic information system to generate an updated scene image until the updated scene image meets a preset iteration condition to obtain a target scene image; and carrying out data processing based on the target scene image to obtain a data processing result.
2. The method according to claim 1, wherein the controlling the visual verification model based on the master model performs hierarchical multi-dimensional image verification on the initial scene image to obtain a structured verification result of the initial scene image, includes: acquiring a natural language instruction input by a user and view state information in the geographic information system, wherein the view state information comprises camera parameters, cone information and figure state information; performing data processing on the natural language instruction and the view state information according to the main control model, and dynamically synthesizing visual verification task description information corresponding to the initial scene image; And carrying out hierarchical multidimensional image verification on the initial scene image and the natural language instruction based on the visual verification large model and the visual verification task description information according to the preset priority corresponding to each image verification dimension to obtain a structured verification result.
3. The method of claim 2, wherein the structured verification result comprises a rendering integrity status result and a semantic consistency result, wherein the performing hierarchical multidimensional image verification on the initial scene image and the natural language instruction based on the visual verification model and the visual verification task description information according to the preset priority corresponding to each image verification dimension to obtain the structured verification result comprises: Performing semantic extraction on the initial scene image and the visual verification task description information based on the visual verification model to obtain visual semantic features corresponding to the initial scene image and text semantic features corresponding to the visual verification task description information; Based on the bottom visual mode in the visual semantic features, identifying whether the initial scene image has rendering abnormality or not, and obtaining the rendering completeness state result; And calculating the semantic matching degree of the initial scene image and the visual verification task description information based on an attention alignment mechanism between the visual semantic features and the text semantic features, and obtaining the semantic consistency result.
4. The method according to claim 1, wherein the controlling the visual verification model based on the master model performs hierarchical multi-dimensional image verification on the initial scene image to obtain a structured verification result of the initial scene image, includes: Carrying out sampling processing on the initial scene image for multiple times according to a preset dynamic sampling strategy and multi-dimensional verification types, and generating to-be-verified images corresponding to the verification types of each dimension respectively; Aiming at the verification type of each hierarchy dimension, carrying out image verification on the image to be verified corresponding to the verification type at present based on the visual verification large model to obtain an initial verification result of the verification type of the current dimension; and generating a structured verification result according to the initial verification result of the verification type of each dimension.
5. The method of claim 1, wherein the structured verification result includes an anomaly priority corresponding to the anomaly information, and the determining, if the structured verification result is that anomaly information exists, a target rectification logic corresponding to the anomaly information in each level of image verification dimensions according to the master control model, the anomaly information and a preset rectification policy includes: And aiming at the structural verification result of each hierarchical image verification dimension, if the structural verification result is that abnormal information exists, matching the abnormal priority with a preset deviation correcting strategy according to the main control model, and determining target deviation correcting logic corresponding to the abnormal information.
6. The method according to claim 5, wherein the determining the target rectification logic corresponding to the anomaly information according to the master control model by matching the anomaly priority and the preset rectification policy if the structured verification result is that anomaly information exists for each hierarchical image verification dimension comprises: If the structural verification result shows that the rendering abnormality exists, determining a target deviation correcting logic as controlling the geographical information system to render again according to the abnormality priority of the rendering abnormality; And if the structured verification result shows that the image content is abnormal, matching a target deviation rectifying logic in the preset deviation rectifying strategy according to the type of the abnormal content according to the abnormal priority of the image content abnormality.
7. A data processing apparatus for application to a geographic agent, the geographic agent comprising a master control model and a visual verification model, the apparatus comprising: The acquisition module is used for acquiring an initial scene image generated by the geographic information system in response to the target operation instruction; The verification module is used for controlling the visual verification large model to carry out multi-dimensional image verification on the initial scene image layer by layer based on the master control model to obtain a structured verification result of the initial scene image; The determining module is used for determining target deviation rectifying logic corresponding to the abnormal information in each level of image verification dimension according to the main control model, the abnormal information and a preset deviation rectifying strategy if the structured verification result is that the abnormal information exists; The deviation rectifying module is used for controlling the geographic information system to generate an updated scene image according to the master control model and the target deviation rectifying logic until the updated scene image meets a preset iteration condition to obtain a target scene image; And the data processing module is used for carrying out data processing based on the target scene image to obtain a data processing result.
8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 6 when the computer program is executed.
9. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.
10. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.

Description

Data processing method, apparatus, computer device and readable storage medium Technical Field The present application relates to the field of geographic information systems and artificial intelligence technologies, and in particular, to a data processing method, apparatus, computer device, and readable storage medium. Background With the development of artificial intelligence agent technology, geographic intelligence agents are presented, which have spatial location awareness, spatial behavior decision-making and spatial interaction capabilities and are capable of automatically executing user commands. In the conventional technology, after a geographic agent receives a natural language instruction input by a user, the geographic agent maps the natural language instruction into a target instruction which can be processed by a GIS (Geographic Information System ) through a master control model (LLM, large Language Model, large language model) in the geographic agent, the GIS updates a current view after executing a task according to the target instruction, feeds back a task completion prompt to the geographic agent after the task is successfully executed, and finally the geographic agent judges that the task is successfully executed based on the completion prompt, thereby realizing data processing such as inspection or monitoring of a target scene. However, in the current conventional technology, the GIS engine is affected by network delay or rendering bottleneck in executing tasks, and the visual image actually presented is often inconsistent with the execution state fed back by the API, so that the geographic agent performs subsequent reasoning based on the wrong execution result, and further the accuracy of the geographic agent data processing is poor. Disclosure of Invention Based on this, it is necessary to provide a data processing method, apparatus, computer device and readable storage medium in order to address the above technical problems. In a first aspect, the present application provides a data processing method, the method being applied to a geographic agent, the geographic agent including a master control model and a visual verification model, comprising: acquiring an initial scene image generated by a geographic information system in response to a target operation instruction; Controlling the visual verification model to perform hierarchical multi-dimensional image verification on the initial scene image based on the master control model to obtain a structured verification result of the initial scene image; If the structured verification result is that abnormal information exists, determining target deviation rectifying logic corresponding to the abnormal information in each level of image verification dimension according to the main control model, the abnormal information and a preset deviation rectifying strategy; According to the master control model and the target deviation rectifying logic, controlling the geographic information system to generate an updated scene image until the updated scene image meets a preset iteration condition to obtain a target scene image; and carrying out data processing based on the target scene image to obtain a data processing result. In one embodiment, the controlling the visual verification large model based on the master control model to perform hierarchical multi-dimensional image verification on the initial scene image to obtain a structured verification result of the initial scene image includes: acquiring a natural language instruction input by a user and view state information in the geographic information system, wherein the view state information comprises camera parameters, cone information and figure state information; performing data processing on the natural language instruction and the view state information according to the main control model, and dynamically synthesizing visual verification task description information corresponding to the initial scene image; And carrying out hierarchical multidimensional image verification on the initial scene image and the natural language instruction based on the visual verification large model and the visual verification task description information according to the preset priority corresponding to each image verification dimension to obtain a structured verification result. In one embodiment, the structured verification result includes a rendering integrity status result and a semantic consistency result, and the performing hierarchical multidimensional image verification on the initial scene image and the natural language instruction based on the visual verification model and the visual verification task description information according to a preset priority corresponding to each image verification dimension to obtain the structured verification result includes: Performing semantic extraction on the initial scene image and the visual verification task description information based on the visual verific