一种近海大区域多尺度浮筏养殖信息智能提取方法及系统
By using a multi-scale floating raft facility annotation dataset and a deep learning model, combined with tidal interference compensation and a multi-level labeling system, the problems of facility identification error and drift in nearshore large-area floating raft aquaculture were solved, achieving accurate facility identification and compliance analysis, and improving the accuracy of aquaculture density statistics and marine resource management capabilities.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- DADI XINYA (BEIJING) TECH CO LTD
- Filing Date
- 2025-08-15
- Publication Date
- 2026-07-17
AI Technical Summary
In large-scale floating raft aquaculture in nearshore areas, satellite remote sensing images are easily affected by tides and waves, leading to feature occlusion, large facility identification errors, undetected facility drift, low identification accuracy in mixed aquaculture areas, inability to achieve hierarchical feature extraction, and affecting the accuracy of aquaculture density statistics.
Using a multi-scale floating raft facility labeled dataset, a deep learning recognition model with multi-scale feature fusion was trained. Combined with tidal interference compensation and a multi-level labeling system, image preprocessing and morphological optimization were performed through a pyramid feature extraction network and a dual-path attention mechanism to generate accurate boundary vector data. Compliance analysis was then conducted in conjunction with a geographic information system.
It improves the completeness and accuracy of identifying targets such as floating rafts, net cages, and ball valves, dynamically assesses the distribution of aquaculture density, generates compliance reports, and enhances the management and control capabilities of marine spatial resources.
Smart Images
Figure CN121053403B_ABST