VOIM: Training-Free Open-Vocabulary 3D Instance Mapping for RGB-D and Monocular SLAM

📄 arXiv: 2609.00775v1 📥 PDF

作者: Sangmin Song, Sarath Kodagoda, Marc G. Carmichael, Karthick Thiyagarajan, Amal Gunatilake, Kelly Prentice, Jodi Martin

分类: cs.CV, cs.AI

发布日期: 2026-09-01

💡 一句话要点

提出VOIM以解决训练无关的开放词汇3D实例映射问题

🎯 匹配领域: 支柱三：空间感知与语义 (Perception & Semantics)

关键词: 3D实例映射 开放词汇 训练无关 RGB-D 单目视觉 体素基础 在线系统 软证据

📋 核心要点

  1. 现有的在线系统在首次检测时就对物体实例进行分割和标记，导致在证据不足时做出决策，影响准确性。
  2. VOIM通过推迟标签和实例决策，利用多视角的软证据来构建更准确的3D实例地图，避免了早期决策的局限性。
  3. 在ScanNet++数据集上，VOIM的mIoU达到了44.07，相较于OVO-SLAM的32.37，显示出显著的性能提升。

📝 摘要（中文）

我们提出了体素基础的在线实例管理器VOIM，它是一种无需训练的体素基础实例管理器，能够从RGB-D或单目RGB构建开放词汇的3D实例地图，这是之前任何训练无关系统未曾解决的领域。VOIM推迟标签和实例决策，直到从未修改的现成感知中积累了每个体素的软证据。实验结果表明，在ScanNet++上，VOIM的映射阶段超越了最强的在线RGB-D系统OVO-SLAM，mIoU提升幅度在4.8到11.7之间。

🔬 方法详解

问题定义：本论文旨在解决现有在线3D实例映射系统在首次检测时做出决策的不足，导致准确性受限的问题。现有方法通常在证据不足时就进行标记，影响最终结果。

核心思路：VOIM的核心思路是推迟标签和实例决策，直到从多个视角收集到足够的软证据，从而提高3D实例映射的准确性和鲁棒性。

技术框架：VOIM的整体架构包括体素分割、软证据积累和实例决策三个主要模块。首先，通过体素化处理输入数据，然后在多个视角下积累每个体素的证据，最后进行实例的标记和决策。

关键创新：VOIM的最大创新在于其训练无关的设计，使其能够在没有预训练模型的情况下，利用现成的感知系统进行高效的3D实例映射。这与现有方法依赖于特定训练模型的方式形成了鲜明对比。

关键设计：在设计中，VOIM采用了多视角证据积累机制，并通过优化的区域描述符和检测器标签先验来提高映射精度。具体的参数设置和损失函数设计未在摘要中详细说明，需参考原文获取更多细节。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

在ScanNet++数据集上，VOIM的mIoU达到了44.07，显著高于OVO-SLAM的32.37，提升幅度在4.8到11.7之间。此外，VOIM在十个场景中均取得胜利，显示出其在开放词汇3D实例映射中的优越性。

🎯 应用场景

该研究的潜在应用领域包括机器人导航、增强现实和智能监控等。通过提供高精度的3D实例地图，VOIM能够在复杂环境中实现更智能的决策和交互，具有重要的实际价值和未来影响。

📄 摘要（原文）

We present Voxel-Grounded Online Instance Manager (VOIM), a training-free voxel-grounded instance manager that builds open-vocabulary 3D instance maps from RGB-D or from monocular RGB alone, a regime no prior training-free system addresses. Online systems typically segment object instances and label them at first detection, committing when evidence is weakest. VOIM instead defers label and instance decisions until soft evidence from unmodified, off-the-shelf perception has accumulated per voxel across views. We show that the mapping stage, rather than the particular perception models, carries the result: across four perception configurations on ScanNet++, varying the region descriptor, the detector label prior and the mask source, the map exceeds the strongest online RGB-D system, OVO-SLAM, by between 4.8 and 11.7 mIoU. Perception is not neutral, and substituting that baseline's own descriptor family costs 4.1 of the margin, yet the baseline carries the marginally better 2D descriptor (33.7 vs. 31.5 mIoU over three scenes) and still realizes the weaker map. Under a like-for-like protocol VOIM reaches 44.07 mIoU on ScanNet++ against 32.37, winning all ten scenes and both aggregations (pooled 33.31 vs. 25.97), and the same system runs unchanged to fully monocular RGB, matching that baseline pooled on Replica (27.80 vs. 27.50). The advantage is regime-specific: under Replica's all-classes scoring, matched inputs give a split result, 28.60 vs. 27.50 pooled against 24.59 vs. 30.11 on the per-scene mean. Room scale is label-limited and building scale drift-limited. Labeling does not run in real time, dominated by per-class detection over the full vocabulary. The maps export occupancy grids and resolve free-form queries to object instances.