Hyperbolic Image-and-Pointcloud Contrastive Learning for 3D Classification

TL;DR

This paper introduces HyperIPC, a hyperbolic space-based contrastive learning method for 3D data that captures hierarchical semantic correlations and improves classification accuracy over Euclidean-based methods.

Contribution

The paper proposes a novel hyperbolic contrastive learning framework for 3D data, enhancing intra-modal feature representation and cross-modal semantic correlation understanding.

Findings

HyperIPC improves object classification accuracy by 2.8% on ScanObjectNN.

HyperIPC enhances few-shot classification results by 5.9%.

Ablation studies confirm the effectiveness of HyperIPC's components.

Abstract

3D contrastive representation learning has exhibited remarkable efficacy across various downstream tasks. However, existing contrastive learning paradigms based on cosine similarity fail to deeply explore the potential intra-modal hierarchical and cross-modal semantic correlations about multi-modal data in Euclidean space. In response, we seek solutions in hyperbolic space and propose a hyperbolic image-and-pointcloud contrastive learning method (HyperIPC). For the intra-modal branch, we rely on the intrinsic geometric structure to explore the hyperbolic embedding representation of point cloud to capture invariant features. For the cross-modal branch, we leverage images to guide the point cloud in establishing strong semantic hierarchical correlations. Empirical experiments underscore the outstanding classification performance of HyperIPC. Notably, HyperIPC enhances object classification results by 2.8% and few-shot classification outcomes by 5.9% on ScanObjectNN compared to the baseline. Furthermore, ablation studies and confirmatory testing validate the rationality of HyperIPC's parameter settings and the effectiveness of its submodules.