English

NeuroVoxel-LM: Language-Aligned 3D Perception via Dynamic Voxelization and Meta-Embedding

Computer Vision and Pattern Recognition 2025-07-29 v1 Artificial Intelligence Machine Learning

Abstract

Recent breakthroughs in Visual Language Models (VLMs) and Multimodal Large Language Models (MLLMs) have significantly advanced 3D scene perception towards language-driven cognition. However, existing 3D language models struggle with sparse, large-scale point clouds due to slow feature extraction and limited representation accuracy. To address these challenges, we propose NeuroVoxel-LM, a novel framework that integrates Neural Radiance Fields (NeRF) with dynamic resolution voxelization and lightweight meta-embedding. Specifically, we introduce a Dynamic Resolution Multiscale Voxelization (DR-MSV) technique that adaptively adjusts voxel granularity based on geometric and structural complexity, reducing computational cost while preserving reconstruction fidelity. In addition, we propose the Token-level Adaptive Pooling for Lightweight Meta-Embedding (TAP-LME) mechanism, which enhances semantic representation through attention-based weighting and residual fusion. Experimental results demonstrate that DR-MSV significantly improves point cloud feature extraction efficiency and accuracy, while TAP-LME outperforms conventional max-pooling in capturing fine-grained semantics from NeRF weights.

Keywords

Cite

@article{arxiv.2507.20110,
  title  = {NeuroVoxel-LM: Language-Aligned 3D Perception via Dynamic Voxelization and Meta-Embedding},
  author = {Shiyu Liu and Lianlei Shan},
  journal= {arXiv preprint arXiv:2507.20110},
  year   = {2025}
}

Comments

**14 pages, 3 figures, 2 tables

R2 v1 2026-07-01T04:20:36.135Z