基于多尺度特征与全局信息交互的光伏组件缺陷检测方法
By constructing a YOLOv8 network and introducing a method that interacts with multi-scale features and global information, the stability and accuracy of photovoltaic module defect detection under different shooting equipment and lighting conditions are solved, improving detection accuracy and adaptability, and making it suitable for real-time detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HEBEI UNIV OF TECH
- Filing Date
- 2025-09-03
- Publication Date
- 2026-07-17
AI Technical Summary
Existing photovoltaic module defect detection methods lack stability and accuracy when processing images from different shooting devices and under different lighting conditions, and are difficult to effectively identify low-contrast and small-sized defect targets.
A photovoltaic module defect detection method based on multi-scale feature and global information interaction is adopted. By constructing a YOLOv8 network and combining low-dimensional and high-dimensional information aggregation and injection modules, a multi-head attention mechanism and a normalized Wasserstein distance loss function are introduced to achieve feature alignment and information interaction, thereby improving detection accuracy and generalization ability.
It significantly improves the accuracy and robustness of photovoltaic module defect detection, maintains high detection performance under different environments, and reduces computational complexity, making it suitable for real-time detection.
Smart Images
Figure CN121147138B_ABST