BioAMASS
收藏资源简介:
BioAMASS数据集旨在为动态人体表面内部的三维骨骼位置和方向的学习提供数据支持。该数据集通过将新的生物力学骨骼模型(BSM)拟合到AMASS数据集中的SMPL人体模型序列中而创建。BioAMASS数据集使得能够训练出一个回归器,将SMPL网格顶点映射到优化的关节位置和骨骼旋转,从而促进了SKEL模型的生成。SKEL模型结合了皮肤和骨骼网格,两者均同步地由相同的姿态参数控制,并且可以通过继承SMPL形状空间进行重塑。
The BioAMASS dataset is developed to provide data support for learning the 3D positions and orientations of skeletal joints within dynamic human body surfaces. It is constructed by fitting a novel biomechanical skeletal model (BSM) to the SMPL human motion sequences from the AMASS dataset. The BioAMASS dataset enables training a regressor that maps SMPL mesh vertices to optimized joint positions and bone rotations, thereby facilitating the generation of the SKEL model. The SKEL model integrates both skin and skeletal meshes, both of which are synchronously controlled by the same pose parameters and can be reshaped by inheriting the SMPL shape space.




