Partition-Invariant Tuning for 3D Scene Understanding

2026-09-11

AI Summary

为解决3D场景理解中因分割变化导致的表示偏移问题，提出PointPiT框架，通过场景感知结构适配器和梯度子空间优化方法提高性能。

Abstract

Scene-level point cloud understanding remains challenging due to diverse geometries and spatial layouts. While pre-trained 3D point cloud foundation models (PFMs) offer strong transferability, full fine-tuning (FFT) incurs substantial computational and storage costs. Parameter-efficient fine-tuning (PEFT) provides a promising alternative, but existing PEFT methods largely focus on object-level point clouds and overlook serialization-induced partition variations in large-scale scenes. To address this issue, we propose PointPiT, a partition-invariant tuning framework for scene-level point clouds. Specifically, a Scene-aware Structural Adapter (SSA) integrates local geometric patterns with global scene context to mitigate partition-induced representation shifts. Moreover, Gradient Subspace Optimization (GSO) selects informative and partition-stable update directions, suppressing partition-dependent variations during optimization. Extensive experiments across multiple scene-level benchmarks demonstrate that PointPiT achieves competitive or even superior performance to full fine-tuning with less than 1% of backbone's parameters, while achieving consistent state-of-the-art performance among representative PEFT methods.

Problem

• 3D Scene Understanding
• Point Clouds
• Fine-tuning
• Parameter Efficiency
• Partition Variations

Innovation

• partition-invariant tuning
• Scene-aware Structural Adapter (SSA)
• Gradient Subspace Optimization (GSO)