传统的植株器官分割方法依赖经验选择阈值参数,而当前的深度学习浅层框架可能会导致植株重要的几何特征丢失,并难以有效整合植株的局部和全局特征。因此,提出了一个基于三维点云的植株器官分割网络(local global feature fusion segment...传统的植株器官分割方法依赖经验选择阈值参数,而当前的深度学习浅层框架可能会导致植株重要的几何特征丢失,并难以有效整合植株的局部和全局特征。因此,提出了一个基于三维点云的植株器官分割网络(local global feature fusion segmentation network,LGF-SegNet)模型,通过引入双权重注意力机制模块和位置编码,更适合在植株点云数据中表达几何特征。在提出的框架的解码层引入特征聚合模块,融合植株点云的局部和全局特征,使得该框架能够关注植株的整体特征轮廓同时保留细节植物纹理(如茎和叶)。实验结果表明,提出的架构在语义分割的交并比、精确率和F1分数的平均值分别达到85.76%、93.18%、91.08%,在实例分割的平均精确率、平均实例覆盖率以及平均实例加权覆盖率达到85.27%、78.46%、79.63%,优于当前流行的植株点云分割任务中使用的深度学习网络架构,并适用于植株语义分割和实例分割的双重任务。这为后续的植株生长预测等研究奠定基础。展开更多
The identification of loci and markers associated with milled grain appearance traits is essential for breeding high-yielding and good-quality rice variety.To detect stable loci for these characteristics,grain length(...The identification of loci and markers associated with milled grain appearance traits is essential for breeding high-yielding and good-quality rice variety.To detect stable loci for these characteristics,grain length(GL),grain width(GW),grain length/width(GLW),chalkiness degree(CD),chalky-grain rate(CR)and translucency degree(TD)of 378 rice lines were evaluated in three seasons.These lines were derived from a multi-parent advanced generation intercross(MAGIC)population.展开更多
基金supported by the earmarked fund for China Agriculture Research System(Grant No.CARS-01)Young Elite Scientists Sponsorship Program by China Association for Science and Technology(Grant No.2020QNRC001)Hunan Academy of Agricultural Sciences Scientific and Technological Innovation Project,China(Grant No.2017JC10)。
文摘The identification of loci and markers associated with milled grain appearance traits is essential for breeding high-yielding and good-quality rice variety.To detect stable loci for these characteristics,grain length(GL),grain width(GW),grain length/width(GLW),chalkiness degree(CD),chalky-grain rate(CR)and translucency degree(TD)of 378 rice lines were evaluated in three seasons.These lines were derived from a multi-parent advanced generation intercross(MAGIC)population.