一种双任务驱动的果梗识别与采摘点定位校正方法及系统
By employing a dual-task driven method for stem identification and picking point localization, and utilizing the StemNet network for stem identification and localization correction, the problem of inaccurate chili picking point localization in complex environments is solved, achieving high-precision and efficient picking point localization, which is suitable for automated agricultural harvesting.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- UNION COLLEGE OF FUJIAN NORMAL UNIV
- Filing Date
- 2025-12-06
- Publication Date
- 2026-07-17
AI Technical Summary
Existing chili picking point positioning models are easily affected by occlusion and the slender shape of the fruit stems in complex environments, resulting in inaccurate picking point positioning and failure of robotic arm picking.
A dual-task driven method for fruit stem recognition and picking point localization is adopted. The StemNet network is used for fruit stem recognition and localization correction. The hybrid attention weighted convolution module HAWConv and the step attention feature fusion module SteAttn are combined to perform fine segmentation of the fruit stem region and spatial correction of the picking point.
It improves the positioning accuracy and robustness of the picking point, reduces the amount of computation, and is suitable for automated agricultural picking scenarios, ensuring the accuracy and real-time nature of the picking process.
Smart Images
Figure CN121661638B_ABST