SurfR: Surface Reconstruction with Multi-scale Attention

TL;DR

SurfR introduces a fast, accurate surface reconstruction method for unorganized point clouds using a multi-scale attention-based implicit representation, balancing speed and detail better than existing approaches.

Contribution

The paper presents a novel implicit representation leveraging multi-scale attention and lazy query techniques, achieving superior speed and comparable accuracy in 3D shape reconstruction.

Findings

Faster reconstruction than baseline methods at optimal resolution

Maintains high surface detail with marginal performance loss

Robust to noise and input resolution variations

Abstract

We propose a fast and accurate surface reconstruction algorithm for unorganized point clouds using an implicit representation. Recent learning methods are either single-object representations with small neural models that allow for high surface details but require per-object training or generalized representations that require larger models and generalize to newer shapes but lack details, and inference is slow. We propose a new implicit representation for general 3D shapes that is faster than all the baselines at their optimum resolution, with only a marginal loss in performance compared to the state-of-the-art. We achieve the best accuracy-speed trade-off using three key contributions. Many implicit methods extract features from the point cloud to classify whether a query point is inside or outside the object. First, to speed up the reconstruction, we show that this feature extraction does not need to use the query point at an early stage (lazy query). Second, we use a parallel multi-scale grid representation to develop robust features for different noise levels and input resolutions. Finally, we show that attention across scales can provide improved reconstruction results.