FisherPoser：基于层级区域 Fisher 矩阵不确定性建模的稀疏观测人体运动估计
FisherPoser: Human Motion Estimation from Sparse Observations with Hierarchical Region-Wise Fisher-Matrix Uncertainty Modeling
将关节旋转建模为 SO(3) 上的矩阵 Fisher 分布，输出每个关节的分布而非单一点估计，为稀疏 VR 追踪等欠约束场景提供可校准的逐关节不确定性；其区域级 token 与层级 Fisher 回归思路可迁移到其他需要概率化旋转输出的姿态估计任务。
原文依据 we introduce a probabilistic framework that models joint orientations as distributions on SO(3) using the Matrix Fisher distribution
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论从稀疏 VR 观测中估计全身运动本质上是一个欠约束问题，仅有三个 6 自由度追踪器（头显与手柄）可用于推断完整骨骼姿态。为缓解这种歧义，我们引入一个概率框架，利用矩阵 Fisher 分布将关节朝向建模为 SO(3) 上的分布。我们的网络不再预测单一确定性姿态，而是为每个关节输出一个分布，其众数与集中度直接量化旋转流形上的预测不确定性。这支持基于似然的训练与有原则的不确定性传播。模型的核心是一个因果 Transformer 编码器，将稀疏观测与运动历史融合。我们进一步提出躯干、手臂和腿部的区域级 token，通过对局部关节特征与语义 VR 锚点进行注意力池化获得。这些 token 引导紧凑的逐区域 Fisher 回归。为高效保证运动学一致性，我们采用肢体细化模块，其中每个子关节的 Fisher 参数以其父关节分布与区域上下文为条件，从而层级式地传播姿态与不确定性。在标准稀疏 VR 基准上的大量实验表明，我们的方法达到最优性能，同时提供良好校准的逐关节不确定性。
英文原文摘要 1364 字符
Full-body motion estimation from sparse VR observations is an inherently under-constrained problem, with only three 6-DoF trackers (HMD and controllers) available to infer a full skeletal pose. To address this ambiguity, we introduce a probabilistic framework that models joint orientations as distributions on SO(3) using the Matrix Fisher distribution. Instead of predicting a single deterministic pose, our network outputs a distribution for each joint, whose mode and concentration directly quantify prediction uncertainty on the rotation manifold. This enables likelihood-based training and principled uncertainty propagation. At the core of our model is a causal Transformer encoder that fuses sparse observations with motion history. We further propose region-wise tokens for the torso, arms, and legs, obtained via attention pooling over local joint features and semantic VR anchors. These tokens guide compact, per-region Fisher regression. To ensure kinematic coherence efficiently, we employ a limb refinement module, where each child joint's Fisher parameters are conditioned on its parent's distribution and the regional context, propagating pose and uncertainty hierarchically. Extensive experiments on standard sparse-VR benchmarks show that our approach achieves state-of-the-art performance, while providing well-calibrated joint-wise uncertainty.