Multimodal Learning 相关度: 9/10

GAP-MLLM: Geometry-Aligned Pre-training for Activating 3D Spatial Perception in Multimodal Large Language Models

Jiaxin Zhang, Junjun Jiang, Haijie Li, Youyu Chen, Kui Jiang, Dave Zhenyu Chen
arXiv: 2603.16461v1 发布: 2026-03-17 更新: 2026-03-17

AI 摘要

GAP-MLLM通过几何对齐预训练,增强MLLM在3D空间感知任务中的性能。

主要贡献

  • 提出GAP-MLLM框架,显式激活MLLM的结构感知能力
  • 引入视觉提示联合任务,预测稀疏点云和语义标签
  • 设计多层渐进融合模块,自适应融合几何先验

方法论

GAP-MLLM通过视觉提示联合任务进行预训练,并使用多层渐进融合模块整合几何信息,提升3D感知能力。

原文摘要

Multimodal Large Language Models (MLLMs) demonstrate exceptional semantic reasoning but struggle with 3D spatial perception when restricted to pure RGB inputs. Despite leveraging implicit geometric priors from 3D reconstruction models, image-based methods still exhibit a notable performance gap compared to methods using explicit 3D data. We argue that this gap does not arise from insufficient geometric priors, but from a misalignment in the training paradigm: text-dominated fine-tuning fails to activate geometric representations within MLLMs. Existing approaches typically resort to naive feature concatenation and optimize directly for downstream tasks without geometry-specific supervision, leading to suboptimal structural utilization. To address this limitation, we propose GAP-MLLM, a Geometry-Aligned Pre-training paradigm that explicitly activates structural perception before downstream adaptation. Specifically, we introduce a visual-prompted joint task that compels the MLLMs to predict sparse pointmaps alongside semantic labels, thereby enforcing geometric awareness. Furthermore, we design a multi-level progressive fusion module with a token-level gating mechanism, enabling adaptive integration of geometric priors without suppressing semantic reasoning. Extensive experiments demonstrate that GAP-MLLM significantly enhances geometric feature fusion and consistently enhances performance across 3D visual grounding, 3D dense captioning, and 3D video object detection tasks.

标签

MLLM 3D Spatial Perception Geometric Alignment Pre-training Visual Grounding

arXiv 分类

cs.CV