G2P: Gaussian-to-Point Attribute Alignment for Boundary-Aware 3D Semantic Segmentation

TL;DR

G2P introduces a novel method that transfers appearance-aware attributes from Gaussian representations to point clouds, enhancing boundary accuracy and discriminative segmentation in 3D scenes without extra supervision.

Contribution

The paper presents Gaussian-to-Point (G2P), a new approach that aligns Gaussian attributes with point clouds to improve 3D semantic segmentation accuracy.

Findings

Achieves superior segmentation performance on benchmarks.

Improves boundary localization in complex scenes.

Enhances discrimination of geometrically similar objects.

Abstract

Semantic segmentation on point clouds is critical for 3D scene understanding. However, sparse and irregular point distributions provide limited appearance evidence, making geometry-only features insufficient to distinguish objects with similar shapes but distinct appearances (e.g., color, texture, material). We propose Gaussian-to-Point (G2P), which transfers appearance-aware attributes from 3D Gaussian Splatting to point clouds for more discriminative and appearance-consistent segmentation. Our G2P address the misalignment between optimized Gaussians and original point geometry by establishing point-wise correspondences. By leveraging Gaussian opacity attributes, we resolve the geometric ambiguity that limits existing models. Additionally, Gaussian scale attributes enable precise boundary localization in complex 3D scenes. Extensive experiments demonstrate that our approach achieves superior performance on standard benchmarks and shows significant improvements on geometrically challenging classes, all without any 2D or language supervision.