S$^2$AM3D：尺度可控的三维点云部件分割
S$^2$AM3D: Scale-controllable Part Segmentation of 3D Point Clouds
把多视图 2D 分割先验通过原生 3D 对比学习聚合成全局一致的点特征，再用连续尺度信号实时调节分割粒度，这套「2D 先验 + 3D 一致性监督 + 尺度提示词解码」的范式可直接迁移到点云部件分割、开放词表 3D 分割等需要跨视图一致性的任务。
原文依据 we propose S^2AM3D, which incorporates 2D segmentation priors with 3D consistent supervision
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论部件级点云分割近期在三维计算机视觉领域受到广泛关注。然而，现有研究受制于两大挑战：原生三维模型因数据稀缺而缺乏泛化能力，而引入二维预训练知识又常常导致不同视图之间的分割结果不一致。为应对这些挑战，我们提出 S^2AM3D，将二维分割先验与三维一致性监督相结合。我们设计了一个点一致部件编码器，通过原生三维对比学习聚合多视图二维特征，生成全局一致的点特征。随后提出尺度感知提示词解码器，借助连续尺度信号实现对分割粒度的实时调节。同时，我们引入一个大规模高质量部件级点云数据集，包含超过 10 万个样本，为模型训练提供充足的监督信号。大量实验表明，S^2AM3D 在多种评测设置下均取得领先性能，在处理复杂结构以及尺寸差异显著的部件时展现出卓越的鲁棒性与可控性。
英文原文摘要 1189 字符
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 S^2AM3D, 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 S^2AM3D achieves leading performance across multiple evaluation settings, exhibiting exceptional robustness and controllability when handling complex structures and parts with significant size variations.