Debiased Multiplex Tokenization Using Mamba-Based Pointers for Efficient and Versatile Map-Free Visual Relocalization
Machine Learning and Knowledge Extraction · 23.03.2026 · 8 (3) · 83
DOI: 10.3390/make8030083 ↗Kurz gefasst
Fachartikel in der Zeitschrift Machine Learning and Knowledge Extraction, online erschienen am 23.03.2026. Verfasst von Wenshuai Wang und 5 weiteren Autor:innen. Die Arbeit nennt 61 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Key Research and Development Program of China, National Natural Science Foundation of China, Guangdong S&T Program.
Abstract (Originalsprache)
Visual localization plays a critical role for mobile robots to estimate their position and orientation in GPS-denied environments. However, its efficiency, robustness, and generalization are fundamentally undermined by severe viewpoint changes and dramatic appearance variations, which present persistent challenges for image-based feature representation and pose estimation under real-world conditions. Recently, map-free visual relocalization (MFVR) has emerged as a promising paradigm for lightweight deployment and privacy isolation on edge devices, while how to learn compact and invariant image tokens without relying on structural 3D maps still remains a core problem, particularly in highly dynamic or long-term scenarios. In this paper, we propose the Debiased Multiplex Tokenizer as a novel method (termed as DMT-Loc) for efficient and versatile MFVR to address these issues. Specifically, DMT-Loc is built upon a pretrained vision Mamba encoder and integrates three key modules for relative pose regression: First, Multiplex Interactive Tokenization yields robust image tokens with non-local affinities and cross-domain descriptions. Second, Debiased Anchor Registration facilitates anchor token matching through proximity graph retrieval and autoregressive pointer attribution. Third, Geometry-Informed Pose Regression empowers multi-layer perceptrons with a symmetric swap gating mechanism operating inside each decoupled regression head to support accurate and flexible pose prediction in both pair-wise and multi-view modes. Extensive evaluations across seven public datasets demonstrate that DMT-Loc substantially outperforms existing baselines and ablation variants in diverse indoor and outdoor environments.
Förderung
- National Key Research and Development Program of China
- National Natural Science Foundation of China
- Guangdong S&T Program
- Major Key Project of Pengcheng Laboratory
- Mobile Information Networks-National Science and Technology Major Project
Zitieren
APA
Wang, W., Liu, H., Li, S., Jiang, P., Che, D., & Ding, R. (2026). Debiased Multiplex Tokenization Using Mamba-Based Pointers for Efficient and Versatile Map-Free Visual Relocalization. Machine Learning and Knowledge Extraction, 8(3), 83. https://doi.org/10.3390/make8030083
Vancouver
Wang W, Liu H, Li S, Jiang P, Che D, Ding R. Debiased Multiplex Tokenization Using Mamba-Based Pointers for Efficient and Versatile Map-Free Visual Relocalization. Machine Learning and Knowledge Extraction. 2026;8(3):83. doi:10.3390/make8030083
BibTeX
@article{Wang2026,
author = {Wang, Wenshuai and Liu, Hong and Li, Shengquan and Jiang, Peifeng and Che, Dandan and Ding, Runwei},
title = {Debiased Multiplex Tokenization Using Mamba-Based Pointers for Efficient and Versatile Map-Free Visual Relocalization},
journal = {Machine Learning and Knowledge Extraction},
year = {2026},
volume = {8},
number = {3},
pages = {83},
doi = {10.3390/make8030083}
}
EUVIMED