Self-Supervised Scene Flow Estimation with Point-Voxel Fusion and Surface Representation

TL;DR

This paper introduces a novel self-supervised scene flow estimation method combining point-voxel fusion and surface representation to improve accuracy and efficiency in capturing 3D motion from point clouds.

Contribution

It proposes a new point-voxel fusion approach with surface feature encoding, outperforming existing self-supervised methods and rivaling fully supervised techniques.

Findings

Outperforms all other self-supervised methods on benchmark datasets.

Achieves significant reduction in endpoint error (EPE) by over 8% on KITTI datasets.

Demonstrates effectiveness of surface features in capturing complex 3D structures.

Abstract

Scene flow estimation aims to generate the 3D motion field of points between two consecutive frames of point clouds, which has wide applications in various fields. Existing point-based methods ignore the irregularity of point clouds and have difficulty capturing long-range dependencies due to the inefficiency of point-level computation. Voxel-based methods suffer from the loss of detail information. In this paper, we propose a point-voxel fusion method, where we utilize a voxel branch based on sparse grid attention and the shifted window strategy to capture long-range dependencies and a point branch to capture fine-grained features to compensate for the information loss in the voxel branch. In addition, since xyz coordinates are difficult to describe the geometric structure of complex 3D objects in the scene, we explicitly encode the local surface information of the point cloud through the umbrella surface feature extraction (USFE) module. We verify the effectiveness of our method by conducting experiments on the Flyingthings3D and KITTI datasets. Our method outperforms all other self-supervised methods and achieves highly competitive results compared to fully supervised methods. We achieve improvements in all metrics, especially EPE, which is reduced by 8.51% on the KITTIo dataset and 10.52% on the KITTIs dataset, respectively.