Smoke detection method, device and equipment based on multi-modal fusion

By synchronously acquiring, spatially calibrating, and dynamically associating multimodal data on a rail-mounted inspection robot, and combining cross-modal feature interaction, the problem of low smoke detection accuracy in existing technologies has been solved, achieving higher detection accuracy and robustness.

CN121884289BActive Publication Date: 2026-06-16GUANGZHOU GUOXUN ROBOT TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
GUANGZHOU GUOXUN ROBOT TECH CO LTD
Filing Date
2026-03-23
Publication Date
2026-06-16

AI Technical Summary

Technical Problem

In existing technologies, smoke detection solutions for rail-mounted inspection robots lack precise spatial alignment and effective feature interaction mechanisms, resulting in poor fusion of image features and point cloud features and low detection accuracy.

Method used

The smoke detection method based on multimodal fusion utilizes a synchronization unit to acquire image data and point cloud data, performs spatial calibration and dynamic correlation, and performs feature fusion through a cross-modal feature interaction mechanism, including the construction of initial pixel point cloud mapping relationship, multi-level feature extraction and cross-modal feature interaction, to generate a fused feature map for smoke detection.

Benefits of technology

It improves the accuracy of smoke detection, reduces false and missed detections, and enhances the reliability and environmental adaptability of the detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121884289B_ABST
    Figure CN121884289B_ABST
Patent Text Reader

Abstract

The application relates to a smoke detection method, device and equipment based on multi-modal fusion. The method controls an acquisition unit to obtain image data and point cloud data of a scene to be monitored through a synchronous unit installed on a hanging rail inspection robot. Based on a pre-established initial pixel point cloud mapping relationship, the image data and the point cloud data are subjected to spatial calibration and dynamic correlation, and the image data and the point cloud data after the spatial calibration and the dynamic correlation are subjected to multi-level feature extraction respectively to obtain an image feature map and a point cloud feature map. Based on a pre-set cross-modal feature interaction mechanism, the image feature map and the point cloud feature map are fused to obtain a fused feature map. The fused feature map is subjected to smoke detection and output of a detection report. The method solves the problem that the prior art lacks accurate spatial alignment and effective feature interaction mechanism between point cloud and image data, and the feature information fusion is poor, and the detection accuracy is low.
Need to check novelty before this filing date? Find Prior Art