A traffic infrastructure 3D multi-modal data generation method and system

By acquiring and integrating multimodal data of 3D BIM components of transportation infrastructure, generating unified text descriptions and storing them, the problem of lack of high-precision three-dimensional multimodal datasets is solved, realizing high-fidelity information fusion of three-dimensional objects and widespread application of virtual space.

CN120994621BActive Publication Date: 2026-01-20CHINA TRANSPORT INFORMATION TECH GRP CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202511517712.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-10-23
Publication Date
2026-01-20
Estimated Expiration
2045-10-23

AI Technical Summary

Technical Problem

The lack of high-precision and high-quality 3D multimodal datasets in existing technologies limits the ability to perceive, understand, and generate 3D physical space in the field of transportation infrastructure.

Method used

By acquiring point cloud modal data, multi-view image modal data, and BIM parameter JSON files of 3D BIM components of transportation infrastructure, a unified text description is generated. Combined with cross-modal text instruction modal data, multiple modal data are integrated and stored to form high-fidelity information.

Benefits of technology

It achieves high-fidelity information fusion of three-dimensional objects of transportation infrastructure, enhances the understanding and generation capabilities of virtual space, and provides rich application potential.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120994621B_ABST
    Figure CN120994621B_ABST
Patent Text Reader

Abstract

The application provides a traffic infrastructure 3D multi-modal data generation method and system, relates to the field of large models and three-dimensional multi-modal technology, and the method comprises the following steps: determining point cloud modal data of a synthetic object based on BIM components of traffic infrastructure, generating text description of the point cloud modal data, determining multi-view image modal data of the synthetic object, generating integrated text description of a plurality of multi-views, acquiring a BIM parameter json file of the synthetic object, generating BIM parameter text description corresponding to the BIM parameter json file, generating unified text description of the synthetic object, generating cross-modal text instruction modal data, integrating and storing the point cloud modal data, the multi-view image modal data, the BIM parameter json file, the unified text description and the cross-modal text instruction modal data according to a preset format, so that the traffic infrastructure three-dimensional object has rich high-fidelity information.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Construction method of multi-modal large-scale scene data set

    CN117437366A

  • Vehicle-based building identification method, device and equipment and storage medium

    CN117542007A