GAPrompt++: Multi-Granular Geometry-Aware Point Cloud Prompt for 3D Vision Model

📄 arXiv: 2609.19716v1 📥 PDF

作者: Zixiang Ai, Zhenyu Cui, Yufei Guo, Wenwen Qiang, Lei Chen, Jiwen Lu, Jiahuan Zhou

分类: cs.CV

发布日期: 2026-09-17

备注: Accepted by TPAMI 2026. Code at https://github.com/PKU-OV3-LAB/GAPromptPlus.git

DOI: 10.1109/TPAMI.2026.3729984

💡 一句话要点

提出GAPrompt++以解决点云适应性不足问题

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

关键词: 点云分析 3D视觉 几何感知 参数高效微调 深度学习

📋 核心要点

  1. 现有的点云适应方法在捕捉几何结构和语义信息方面存在不足，导致适应能力受限。
  2. GAPrompt++通过引入多粒度几何特征提取和提示传播机制，增强了模型对几何信息的捕捉能力。
  3. 实验结果显示，GAPrompt++在多个基准测试中超越了全量微调，且仅需不到2%的可训练参数。

📝 摘要（中文）

预训练的3D视觉模型在点云分析中取得了显著进展，但通过全量微调适应下游任务的计算成本和存储需求较高。参数高效微调（PEFT）为此提供了有前景的替代方案。然而，现有的基于提示的方法忽视了点云的内在几何结构，限制了其适应能力。为了解决这些挑战，本文提出了GAPrompt++，一种多粒度几何感知的提示方法，提供了更丰富的几何指导以实现高效的3D任务适应。实验表明，GAPrompt++在多项基准测试中实现了最先进的性能，且所需可训练参数不足2%。

🔬 方法详解

问题定义：本文旨在解决现有点云适应方法在几何结构编码和信息传播方面的不足，导致适应能力受限。

核心思路：GAPrompt++通过多粒度几何特征提取和提示传播机制，提供更丰富的几何指导，从而提高3D任务的适应效率。

技术框架：整体架构包括点位移提示器、关键点提示器和提示传播机制，分别负责提取多粒度几何特征、生成局部几何提示和在特征提取层次中传播几何信息。

关键创新：最重要的创新点在于引入了多粒度几何特征提取和提示传播机制，显著增强了模型对几何特征的捕捉能力，与传统方法相比，能够更有效地适应不同的3D任务。

关键设计：关键设计包括点位移提示器的多尺度特征提取策略、关键点提示器的自适应生成机制，以及在特征提取层次中有效传播几何提示的机制。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

GAPrompt++在多个基准测试中表现优异，超越了现有的提示基础PEFT方法，甚至在全量微调上也取得了更好的结果。具体而言，其在某些任务上性能提升幅度超过了10%，且训练参数量不足2%。

🎯 应用场景

GAPrompt++的研究成果在自动驾驶、机器人导航、虚拟现实等领域具有广泛的应用潜力。通过提高3D视觉模型的适应性，该方法能够更好地处理复杂的点云数据，推动相关技术的发展和应用。

📄 摘要（原文）

Pre-trained 3D vision models have substantially advanced point cloud analysis, yet adapting them to downstream tasks via full fine-tuning is computationally expensive and storage-intensive. Parameter-Efficient Fine-Tuning (PEFT) offers a promising alternative by reducing both adaptation cost and storage burden. However, existing prompting-based approaches ignore the intrinsic geometric structures of point clouds, thereby limiting their adaptation capability. This limitation stems from their inability to encode both fine-grained geometric cues and coarse-grained structural semantics, as well as failing to propagate such information effectively through the model hierarchy. To address these challenges, we propose GAPrompt++, a multi-granular geometry-aware prompting method that provides richer geometric guidance for efficient 3D task adaptation. Specifically, we introduce a Point Shift Prompter that extracts multi-granular geometric features across different scales, enabling instance-specific geometric adjustments during adaptation. Next, a Keypoint Prompter adaptively generates point-level prompts to highlight local geometric saliency and fine-grained structural details. Furthermore, a Prompt Propagation mechanism injects these multi-granular geometric cues throughout the feature extraction hierarchy, strengthening the ability to capture essential geometric characteristics. Extensive experiments show that GAPrompt++ achieves state-of-the-art performance among prompting-based PEFT methods and even surpasses full fine-tuning across diverse benchmarks, while requiring less than 2\% trainable parameters. In addition, to address the saturation of existing evaluation datasets, we construct two more challenging benchmarks derived from 3D Gaussian Splatting and Multi-View Stereo reconstruction, offering diverse and realistic point cloud scenarios to promote future research.