S2AM3D: Scale-controllable Part Segmentation of 3D Point Clouds

TL;DR

S2AM3D introduces a novel method for 3D point cloud segmentation that combines 2D priors with 3D supervision, enabling scalable, consistent, and controllable part segmentation with a new large dataset.

Contribution

The paper presents a scale-aware, view-consistent segmentation framework and a large-scale dataset, advancing 3D point cloud part segmentation with improved robustness and flexibility.

Findings

Achieves state-of-the-art performance on multiple benchmarks.

Demonstrates robust segmentation across complex structures and size variations.

Enables real-time control of segmentation granularity.

Abstract

Part-level point cloud segmentation has recently attracted significant attention in 3D computer vision. Nevertheless, existing research is constrained by two major challenges: native 3D models lack generalization due to data scarcity, while introducing 2D pre-trained knowledge often leads to inconsistent segmentation results across different views. To address these challenges, we propose S2AM3D, which incorporates 2D segmentation priors with 3D consistent supervision. We design a point-consistent part encoder that aggregates multi-view 2D features through native 3D contrastive learning, producing globally consistent point features. A scale-aware prompt decoder is then proposed to enable real-time adjustment of segmentation granularity via continuous scale signals. Simultaneously, we introduce a large-scale, high-quality part-level point cloud dataset with more than 100k samples, providing ample supervision signals for model training. Extensive experiments demonstrate that S2AM3D achieves leading performance across multiple evaluation settings, exhibiting exceptional robustness and controllability when handling complex structures and parts with significant size variations.