Beyond First-Order: Learning Riemannian Geometries for Invariant Visual Place Recognition

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Cheng, Jintao, Li, Weibin, He, Zhijian, Wu, Jin, Vong, Chi Man, Zhang, Wei
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866913131486248960
author Cheng, Jintao
Li, Weibin
He, Zhijian
Wu, Jin
Vong, Chi Man
Zhang, Wei
author_facet Cheng, Jintao
Li, Weibin
He, Zhijian
Wu, Jin
Vong, Chi Man
Zhang, Wei
contents Visual Place Recognition (VPR) demands representations robust to drastic environmental and viewpoint shifts. Existing aggregation paradigms either depend on extensive supervised training or rely on first-order pooling, often struggling to preserve structural correlations under extreme shifts or incurring high adaptation costs. In this work, we propose Riemannian Invariant Aggregation (RIA), a unified geometric framework that explicitly models second-order scene structure on the Symmetric Positive Definite (SPD) manifold. By treating perturbations as tractable congruence transformations, RIA leverages geometry-aware Riemannian mappings to project covariance descriptors into a linearized Euclidean space, effectively preserving invariant structural components while suppressing noise. Extensive evaluations demonstrate that RIA achieves zero-shot performance comparable to supervised methods, and establishes state-of-the-art accuracy with simple fine-tuning, particularly in unstructured environments. The source code will be released.
format Preprint
id arxiv_https___arxiv_org_abs_2602_00841
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Beyond First-Order: Learning Riemannian Geometries for Invariant Visual Place Recognition
Cheng, Jintao
Li, Weibin
He, Zhijian
Wu, Jin
Vong, Chi Man
Zhang, Wei
Computer Vision and Pattern Recognition
Visual Place Recognition (VPR) demands representations robust to drastic environmental and viewpoint shifts. Existing aggregation paradigms either depend on extensive supervised training or rely on first-order pooling, often struggling to preserve structural correlations under extreme shifts or incurring high adaptation costs. In this work, we propose Riemannian Invariant Aggregation (RIA), a unified geometric framework that explicitly models second-order scene structure on the Symmetric Positive Definite (SPD) manifold. By treating perturbations as tractable congruence transformations, RIA leverages geometry-aware Riemannian mappings to project covariance descriptors into a linearized Euclidean space, effectively preserving invariant structural components while suppressing noise. Extensive evaluations demonstrate that RIA achieves zero-shot performance comparable to supervised methods, and establishes state-of-the-art accuracy with simple fine-tuning, particularly in unstructured environments. The source code will be released.
title Beyond First-Order: Learning Riemannian Geometries for Invariant Visual Place Recognition
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2602.00841