Joint clearance,as an important stochastic factor,can significantly deteriorate positioning and repeatability accuracies and lower assembly quality of a 6-DOF docking mechanism.Considering pose accuracy with tradition...Joint clearance,as an important stochastic factor,can significantly deteriorate positioning and repeatability accuracies and lower assembly quality of a 6-DOF docking mechanism.Considering pose accuracy with traditional error model that possesses inherent imprecision,both probabilistic and deterministic approaches based on forward kinematics are presented to analyze comprehensive pose error(CPE)in simulation.Results indicate an identical trend emerges for each CPE with both approaches,and both CPEs perform opposite variations as the moving platform upgrades.The findings provide theoretical reference for refinement of assembly quality evaluation of this mechanism.展开更多
人体姿态估计是计算机视觉的基础性算法之一,为了探究人体姿态估计领域的研究发展趋势,文章首先介绍了基于卷积的经典人体姿态估计算法,论述各算法的基本原理及算法改进,其次对最新的基于自注意力模型(Transformer)的算法进行梳理,最后...人体姿态估计是计算机视觉的基础性算法之一,为了探究人体姿态估计领域的研究发展趋势,文章首先介绍了基于卷积的经典人体姿态估计算法,论述各算法的基本原理及算法改进,其次对最新的基于自注意力模型(Transformer)的算法进行梳理,最后介绍了常用的公开数据集和模型评价指标,选取了几个经典算法进行对比分析,平均精度在马克斯·普朗克信息研究所(Max Planck Institute Informatik,MPII)数据集达到80%以上,在微软公共对象上下文(Common Objects in Context,COCO)数据集达到60%以上,得到卷积结构和Transformer结构互有优劣的结论。展开更多
基金supported by the National Defense Basic Scientific Research Program(No.A0320110019)the Shanghai Science and Technology Innovation Action Plan(No.11DZ1120800)
文摘Joint clearance,as an important stochastic factor,can significantly deteriorate positioning and repeatability accuracies and lower assembly quality of a 6-DOF docking mechanism.Considering pose accuracy with traditional error model that possesses inherent imprecision,both probabilistic and deterministic approaches based on forward kinematics are presented to analyze comprehensive pose error(CPE)in simulation.Results indicate an identical trend emerges for each CPE with both approaches,and both CPEs perform opposite variations as the moving platform upgrades.The findings provide theoretical reference for refinement of assembly quality evaluation of this mechanism.
文摘人体姿态估计是计算机视觉的基础性算法之一,为了探究人体姿态估计领域的研究发展趋势,文章首先介绍了基于卷积的经典人体姿态估计算法,论述各算法的基本原理及算法改进,其次对最新的基于自注意力模型(Transformer)的算法进行梳理,最后介绍了常用的公开数据集和模型评价指标,选取了几个经典算法进行对比分析,平均精度在马克斯·普朗克信息研究所(Max Planck Institute Informatik,MPII)数据集达到80%以上,在微软公共对象上下文(Common Objects in Context,COCO)数据集达到60%以上,得到卷积结构和Transformer结构互有优劣的结论。