HypeVPR: Exploring Hyperbolic Space for Perspective to Equirectangular Visual Place Recognition

2025-06-05

AI Summary

This work addresses the cross-view visual place recognition (VPR) problem between pinhole and equirectangular (P2E) images in mobile robotics. To tackle the semantic and geometric mismatch across these distinct projection models, we propose the first hyperbolic-space-based hierarchical embedding framework. Methodologically, we leverage the intrinsic capability of hyperbolic geometry to model hierarchical scene structures, designing a local-to-global hyperbolic feature aggregation mechanism. We further integrate cross-modal contrastive learning with a coarse-to-fine nearest-neighbor search strategy to achieve efficient and accurate P2E matching. Our key contribution is the novel introduction of hyperbolic embeddings into VPR—overcoming the fundamental limitation of Euclidean representations in capturing scene-level semantic hierarchies. Extensive experiments on multiple standard benchmarks demonstrate significant improvements over state-of-the-art methods: retrieval speed increases by up to 5× while maintaining higher accuracy.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multi-instance/Multi-view Learning

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

When applying Visual Place Recognition (VPR) to real-world mobile robots and similar applications, perspective-to-equirectangular (P2E) formulation naturally emerges as a suitable approach to accommodate diverse query images captured from various viewpoints. In this paper, we introduce HypeVPR, a novel hierarchical embedding framework in hyperbolic space, designed to address the unique challenges of P2E VPR. The key idea behind HypeVPR is that visual environments captured by panoramic views exhibit inherent hierarchical structures. To leverage this property, we employ hyperbolic space to represent hierarchical feature relationships and preserve distance properties within the feature space. To achieve this, we propose a hierarchical feature aggregation mechanism that organizes local-to-global feature representations within hyperbolic space. Additionally, HypeVPR adopts an efficient coarse-to-fine search strategy, optimally balancing speed and accuracy to ensure robust matching, even between descriptors from different image types. This approach enables HypeVPR to outperform state-of-the-art methods while significantly reducing retrieval time, achieving up to 5x faster retrieval across diverse benchmark datasets. The code and models will be released at https://github.com/suhan-woo/HypeVPR.git.

Problem

• Addressing perspective-to-equirectangular challenges in Visual Place Recognition
• Leveraging hyperbolic space for hierarchical feature representation in VPR
• Balancing speed and accuracy in cross-image type descriptor matching

Innovation

• Hierarchical embedding in hyperbolic space
• Coarse-to-fine search strategy
• Local-to-global feature aggregation