Vision-language models like CLIP can offer a promising foundation for 3D scene understanding when extended with 3D tokenizers. However, standard approaches, such as k-nearest neighbor or radius-based tokenization, struggle with cross-domain generalization due to sensitivity to dataset-specific spatial scales. We present a universal 3D tokenizer designed for scale-invariant representation learning with a frozen CLIP backbone. We show that combining superpoint-based grouping with coordinate scale normalization consistently outperforms conventional methods through extensive experimental analysis. Specifically, we introduce S4Token, a tokenization pipeline that produces semantically-informed tokens regardless of scene scale. Our tokenizer is trained without annotations using masked point modeling and clustering-based objectives, along with cross-modal distillation to align 3D tokens with 2D multi-view image features. For dense prediction tasks, we propose a superpoint-level feature propagation module to recover point-level detail from sparse tokens.
@article{arxiv.2505.18819,
title = {Self-Supervised and Generalizable Tokenization for CLIP-Based 3D Understanding},
author = {Guofeng Mei and Bin Ren and Juan Liu and Luigi Riz and Xiaoshui Huang and Xu Zheng and Yongshun Gong and Ming-Hsuan Yang and Nicu Sebe and Fabio Poiesi},
journal= {arXiv preprint arXiv:2505.18819},
year = {2025}
}