An electrical fire detection method and system in a multi-scale complex background
By improving the dynamic feature extraction and feature fusion methods of the RT-DETR model, the problem of insufficient multi-scale accuracy in electrical fire detection under complex backgrounds was solved, and the accurate identification and location of fire targets were achieved.
CN122157158APending Publication Date: 2026-06-05CHINA UNIV OF MINING & TECH
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA UNIV OF MINING & TECH
- Filing Date
- 2026-03-05
- Publication Date
- 2026-06-05
Smart Images
Figure CN122157158A_ABST
Abstract
The application discloses an electrical fire detection method and system in a multi-scale complex background, and belongs to the technical field of image processing. The method comprises the following steps: acquiring an electrical fire picture to be detected; inputting electrical fire picture data into an improved backbone network to generate multi-layer feature data corresponding to different scales respectively; inputting the multi-layer feature data into a hybrid encoder to generate fused feature data; and inputting the fused feature data into a decoder to generate electrical fire detection result data containing boundary box data and confidence score data. The application solves the problems of incomplete feature extraction and insufficient target positioning accuracy of the fire detection model in the prior art in a complex scene.
Need to check novelty before this filing date? Find Prior Art