A power distribution component overheat detection method and system fusing visual and textual information, medium and product
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-20
- Publication Date
- 2026-07-17
AI Technical Summary
Existing infrared overheat detection technology for UAV power distribution components suffers from problems such as blurred target boundaries, high false detection and false negative rates, and lack of understanding of the semantic structure of power equipment and adaptability to complex environments in complex backgrounds, low light, and temperature fluctuation environments.
The detection method integrates visual and textual information. By introducing an instance segmentation model and a large language model, a segmentation mask is generated, temperature features are extracted and normalized, and the total score is calculated by combining the overheating probability, so as to achieve accurate overheating determination of power distribution components.
It significantly reduces the false detection and false negative rates, improves the recognition accuracy and adaptability in complex environments, and ensures the accuracy and reliability of overheat detection.
Smart Images
Figure CN122416019A_ABST