RMS-FlowNet: Efficient and Robust Multi-Scale Scene Flow Estimation for Large-Scale Point Clouds

TL;DR

RMS-FlowNet is a new end-to-end neural network architecture that efficiently estimates scene flow in high-density point clouds using a novel random sampling approach, outperforming existing methods in speed and accuracy.

Contribution

It introduces a fully supervised, multi-scale scene flow estimation method based on random sampling, improving robustness and efficiency over structure-based sampling techniques.

Findings

Achieves higher accuracy than state-of-the-art methods on FlyingThings3D.

Operates efficiently on dense point clouds with over 250K points.

Demonstrates competitive generalization to real-world KITTI data without fine-tuning.

Abstract

The proposed RMS-FlowNet is a novel end-to-end learning-based architecture for accurate and efficient scene flow estimation which can operate on point clouds of high density. For hierarchical scene flow estimation, the existing methods depend on either expensive Farthest-Point-Sampling (FPS) or structure-based scaling which decrease their ability to handle a large number of points. Unlike these methods, we base our fully supervised architecture on Random-Sampling (RS) for multiscale scene flow prediction. To this end, we propose a novel flow embedding design which can predict more robust scene flow in conjunction with RS. Exhibiting high accuracy, our RMS-FlowNet provides a faster prediction than state-of-the-art methods and works efficiently on consecutive dense point clouds of more than 250K points at once. Our comprehensive experiments verify the accuracy of RMS-FlowNet on the established FlyingThings3D data set with different point cloud densities and validate our design choices. Additionally, we show that our model presents a competitive ability to generalize towards the real-world scenes of KITTI data set without fine-tuning.