This paper proposes a method to recognize human-object interactions by modeling context between human actions and interacted objects.Human-object interaction recognition is a challenging task due to severe occlusion b...This paper proposes a method to recognize human-object interactions by modeling context between human actions and interacted objects.Human-object interaction recognition is a challenging task due to severe occlusion between human and objects during the interacting process.Since that human actions and interacted objects provide strong context information,i.e.some actions are usually related to some specific objects,the accuracy of recognition is significantly improved for both of them.Through the proposed method,both global and local temporal features from skeleton sequences are extracted to model human actions.In the meantime,kernel features are utilized to describe interacted objects.Finally,all possible solutions from actions and objects are optimized by modeling the context between them.The results of experiments demonstrate the effectiveness of our method.展开更多
At a conference in Beijing on December 4, 2012 marking the 30th anni- versary of the adoption of the current Constitution, General Secretary of the CPC Central Committee Xi Jinping pointed out, ＂To fully implement th...At a conference in Beijing on December 4, 2012 marking the 30th anni- versary of the adoption of the current Constitution, General Secretary of the CPC Central Committee Xi Jinping pointed out, ＂To fully implement the Constitution is the primary task and the basic work in building a socialist nation ruled by law.＂ He also said, ＂A country ruled by law should be first ruled by the Constitution, and lawful governance should be based on the Constitution.＂展开更多
文摘This paper proposes a method to recognize human-object interactions by modeling context between human actions and interacted objects.Human-object interaction recognition is a challenging task due to severe occlusion between human and objects during the interacting process.Since that human actions and interacted objects provide strong context information,i.e.some actions are usually related to some specific objects,the accuracy of recognition is significantly improved for both of them.Through the proposed method,both global and local temporal features from skeleton sequences are extracted to model human actions.In the meantime,kernel features are utilized to describe interacted objects.Finally,all possible solutions from actions and objects are optimized by modeling the context between them.The results of experiments demonstrate the effectiveness of our method.
文摘At a conference in Beijing on December 4, 2012 marking the 30th anni- versary of the adoption of the current Constitution, General Secretary of the CPC Central Committee Xi Jinping pointed out, ＂To fully implement the Constitution is the primary task and the basic work in building a socialist nation ruled by law.＂ He also said, ＂A country ruled by law should be first ruled by the Constitution, and lawful governance should be based on the Constitution.＂
基金SuppoSed by the National Natural Science Foundation of China under Grant Nos.6067319560703078(国家自然科学基金)+2 种基金the National High-Tech Research and Development Plan of China under Grant No.2007AA04Z113(国家高技术研究发展计划(863))the National Basic Research Program of China under Grant No.2006CB303105(国家重点基础研究发展规划(973))the National Key Technology R&D Program of China under Grant No.2006BAF01A17(国家科技支撑计划)
文摘深度歧义是单帧图像多人3D姿态估计面临的重要挑战,提取图像上下文对缓解深度歧义极具潜力.自顶向下方法大多基于人体检测建模关键点关系,人体包围框粒度粗背景噪声占比较大,极易导致关键点偏移或误匹配,还将影响基于人体尺度因子估计绝对深度的可靠性.自底向上的方法直接检出图像中的人体关键点再逐一恢复3D人体姿态.虽然能够显式获取场景上下文,但在相对深度估计方面处于劣势.提出新的双分支网络,自顶向下分支基于关键点区域提议提取人体上下文,自底向上分支基于三维空间提取场景上下文.提出带噪声抑制的人体上下文提取方法,通过建模“关键点区域提议”描述人体目标,建模姿态关联的动态稀疏关键点关系剔除弱连接减少噪声传播.提出从鸟瞰视角提取场景上下文的方法,通过建模图像深度特征并映射鸟瞰平面获得三维空间人体位置布局;设计人体和场景上下文融合网络预测人体绝对深度.在公开数据集MuPoTS-3D和Human3.6M上的实验结果表明:与同类先进模型相比,所提模型HSC-Pose的相对和绝对3D关键点位置精度至少提高2.2%和0.5%;平均根关键点位置误差至少降低4.2 mm.
