[论文] Cepheid: Decoupling Scale from Structure for Cross-Family Motion Compl...

论文概要

研究领域: CV
作者: Chih-Hai Su, Tsung-Wei Ke, Yu-Lun Liu
发布时间: 2026-10-13
arXiv: 2610.18871

中文摘要

稀疏到稠密运动补全旨在从稀疏顶点位移恢复稠密网格运动。在单一几何族上训练的残差先验在跨族场景中往往无法击败无参数的调和扩展方法。跨族训练还面临缩放障碍：残差幅值在不同几何族之间变化超过一个数量级，且随观测变稠密而坍缩。由于该尺度取决于未知的目标运动，测试时无法直接测量。我们通过学习无量纲先验将残差结构与物理尺度解耦，从可观测量估计幅值，并仅在训练源上拟合轻量级校正。我们的统一模型在所有评估族（包括未见过的布料和有腿生物体）上超越调和扩展，无需骨架、配准或跨网格对应。同一先验无需重训练即可应用于点追踪动物视频和机器人抓取软体。

原文摘要

Sparse-to-dense motion completion recovers dense mesh motion from sparse vertex displacements. Residual priors trained on one geometry family frequently fail to beat parameter-free harmonic extension off-family. Moreover, cross-family training meets a scaling barrier: residual amplitude varies by more than an order of magnitude across geometry families and collapses as observations become denser. Since this scale depends on unknown target motion, it cannot be measured directly at test time. We decouple residual structure from physical scale by learning a dimensionless prior, estimating amplitude from observables and a lightweight correction fitted only on training sources. Our unified model surpasses harmonic extension across all evaluated families, including unseen cloth and legged bodies...

自动采集于 2026-10-15

#论文 #arXiv #CV #小凯