一种基于图像识别的鱼群摄食节律精准投喂方法
By constructing a rhythmic TAPIR model, extracting fish body, bait, and disturbance query points, matching frame by frame to generate linked trajectories, calculating feeding phase states, identifying effective feeding results, and generating precise feeding control parameters, the problem of feeding rhythms being difficult to accurately reflect in existing feeding methods is solved, and precise feeding control is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- FUJIAN GANDONG MARITIME GRANARY BREEDING CO LTD
- Filing Date
- 2026-06-17
- Publication Date
- 2026-07-17
AI Technical Summary
Existing feeding methods are difficult to accurately reflect the actual feeding rhythm of fish at different feeding stages, resulting in underfeeding, overfeeding and uneaten food accumulation. Furthermore, image recognition-based methods are prone to misjudging the feeding status.
A rhythmic TAPIR model is constructed. The fish body, bait and disturbance query points are extracted through the heterogeneous query unit. The linkage trajectory is generated by frame-by-frame matching. The phase refinement unit calculates the feeding phase state and the event decomposition unit identifies the effective feeding results to generate precise feeding control parameters.
It improves the accuracy of judging the feeding status of fish, avoids deviations in judging feeding rhythm, and enables precise feeding control based on changes in feeding rhythm, reducing overfeeding, missed feeding, and misfeeding.
Smart Images
Figure CN122397664A_ABST