Priors for people tracking from small training sets
Priors for people tracking from small training sets
复制标题
DOI：
10.1109/iccv.2005.193
复制
发表时间：
2005-10
期刊：
影响因子：
--
通讯作者：
R. Urtasun;David J. Fleet;Aaron Hertzmann;P. Fua
中科院分区：
文献类型：
--
作者：
R. Urtasun;David J. Fleet;Aaron Hertzmann;P. Fua
我们提倡使用缩放高斯过程潜变量模型（SGPLVM）来学习三维人体姿态的先验模型，以实现三维人体跟踪。SGPLVM同时优化了高维姿态数据的低维嵌入和密度函数，这两者都为接近训练数据的点提供了更高的概率，并提供了从低维潜在空间到全维姿态空间的非线性概率映射。当只有少量训练数据可用时，SGPLVM是一个自然的选择。我们用两种不同的运动来展示我们的方法，打高尔夫球和走路。我们表明，SGPLVM充分约束的问题，跟踪可以完成简单的确定性优化。
We advocate the use of scaled Gaussian process latent variable models (SGPLVM) to learn prior models of 3D human pose for 3D people tracking. The SGPLVM simultaneously optimizes a low-dimensional embedding of the high-dimensional pose data and a density function that both gives higher probability to points close to training data and provides a nonlinear probabilistic mapping from the low-dimensional latent space to the full-dimensional pose space. The SGPLVM is a natural choice when only small amounts of training data are available. We demonstrate our approach with two distinct motions, golfing and walking. We show that the SGPLVM sufficiently constrains the problem such that tracking can be accomplished with straightforward deterministic optimization.