arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2506.05282cs.CVcs.AIcs.RO
收藏

整流点流:通用点云姿态估计

Rectified Point Flow: Generic Point Cloud Pose Estimation

Stanford University(斯坦福大学);NVIDIA Research(NVIDIA研究)

Tao Sun, Liyuan Zhu, Shengyu Huang, Shuran Song, Iro Armeni

AI总结:

提出整流点流,一种统一条件生成框架,通过学习点级速度场实现点云配准与形状装配,无需对称标签,在六个基准上达到最先进性能。

摘要:

我们提出了整流点流(Rectified Point Flow),一种统一的参数化方法,将成对点云配准和多部件形状装配表述为单一的条件生成问题。给定未设置姿态的点云,我们的方法学习一个连续的点级速度场,将噪声点传输到其目标位置,从而恢复部件姿态。与先前采用临时对称性处理的回归部件姿态的工作不同,我们的方法在无对称标签的情况下内在地学习装配对称性。结合一个专注于重叠点的自监督编码器,我们的方法在涵盖成对配准和形状装配的六个基准上取得了新的最先进性能。值得注意的是,我们的统一表述能够有效联合训练不同数据集,促进共享几何先验的学习,从而提升精度。项目页面:https://rectified-pointflow.github.io/。

英文摘要:

We introduce Rectified Point Flow, a unified parameterization that formulates pairwise point cloud registration and multi-part shape assembly as a single conditional generative problem. Given unposed point clouds, our method learns a continuous point-wise velocity field that transports noisy points toward their target positions, from which part poses are recovered. In contrast to prior work that regresses part-wise poses with ad-hoc symmetry handling, our method intrinsically learns assembly symmetries without symmetry labels. Together with a self-supervised encoder focused on overlapping points, our method achieves a new state-of-the-art performance on six benchmarks spanning pairwise registration and shape assembly. Notably, our unified formulation enables effective joint training on diverse datasets, facilitating the learning of shared geometric priors and consequently boosting accuracy. Project page: https://rectified-pointflow.github.io/.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑