一种船舶检测方法、装置、介质和设备
By improving the backbone network and loss function of the YOLOv7 model, the accuracy problem in complex terrain and multi-scale ship detection was solved, achieving a more efficient ship recognition effect.
CN120032250BActive Publication Date: 2026-07-17WUXI UNIV
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- WUXI UNIV
- Filing Date
- 2025-01-26
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing technologies suffer from low detection accuracy in complex terrain and multi-scale ship inspection, especially in environments such as ports and islands where the detection accuracy is insufficient when ships of various scales obstruct the view.
Method used
An improved YOLOv7 model is adopted, replacing the ELAN module in the backbone network with the RFEFM module. Combined with the HLF-FPN network and the F-MPDIoU loss function, feature extraction and feature fusion are enhanced, thereby improving the model's scale adaptability and detection accuracy.
Benefits of technology
It significantly improves the accuracy of ship detection, reduces false detections and missed detections, and enhances the model's ability to identify complex backgrounds and multi-scale targets.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN120032250B_ABST
Abstract
本发明公开了一种船舶检测方法、装置、介质和设备，涉及目标检测技术领域，包括：获取船舶的遥感图像，构建船舶图像数据集；构建改进型YOLOv7模型；基于原有的YOLOv7模型，将原有主干网络中的ELAN模块替换为感受野增强特征提取RFEFM模块，对船舶图像数据集中的不同尺度的船舶特征进行提取，在原有PAFPN特征金字塔结构的基础上，通过增加高低维度特征融合模块HLF模块，得到高低维融合特征金字塔HLF‑FPN网络，进行特征融合；对改进型YOLOv7模型进行训练，获取用于船舶检测的船舶检测模型；获取待识别的船舶图像，将船舶图像输入至船舶检测模型进行船舶识别，能够准确获取船舶检测结果。
Need to check novelty before this filing date? Find Prior Art