PCSTracker: Long-Term Scene Flow Estimation for Point Cloud Sequences

TL;DR

PCSTracker is an innovative end-to-end framework for consistent long-term scene flow estimation in point cloud sequences, addressing challenges like geometric evolution and occlusions with novel modules and sliding-window inference.

Contribution

It introduces the first end-to-end method with iterative optimization and spatio-temporal modules for stable long-term scene flow estimation in point clouds.

Findings

Achieves state-of-the-art accuracy on PointOdyssey3D and ADT3D datasets.

Maintains real-time performance at 32.5 FPS.

Outperforms RGB-D-based approaches in 3D motion understanding.

Abstract

Point cloud scene flow estimation is fundamental to long-term and fine-grained 3D motion analysis. However, existing methods are typically limited to pairwise settings and struggle to maintain temporal consistency over long sequences as geometry evolves, occlusions emerge, and errors accumulate. In this work, we propose PCSTracker, the first end-to-end framework specifically designed for consistent scene flow estimation in point cloud sequences. Specifically, we introduce an iterative geometry motion joint optimization module (IGMO) that explicitly models the temporal evolution of point features to alleviate correspondence inconsistencies caused by dynamic geometric changes. In addition, a spatio-temporal point trajectory update module (STTU) is proposed to leverage broad temporal context to infer plausible positions for occluded points, ensuring coherent motion estimation. To further handle long sequences, we employ an overlapping sliding-window inference strategy that alternates cross-window propagation and in-window refinement, effectively suppressing error accumulation and maintaining stable long-term motion consistency. Extensive experiments on the synthetic PointOdyssey3D and real-world ADT3D datasets show that PCSTracker achieves the best accuracy in long-term scene flow estimation and maintains real-time performance at 32.5 FPS, while demonstrating superior 3D motion understanding compared to RGB-D-based approaches.