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

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.

Authors

• Naiwen Hu
• Haozhe Cheng
• Yifan Xie
• Pengcheng Shi
• Jihua Zhu

Keywords

• Point cloud compression
• Representation learning
• Image segmentation
• Three-dimensional displays
• Correlation
• Semantics
• Contrastive learning
• Space exploration
• Intelligent robots
• Testing
• Self-supervised Learning
• 3D Classification
• Point Cloud
• Object Classification
• Geometric Structure
• Euclidean Space
• Submodule
• Point Cloud Representation
• Few-shot Classification
• Hyperbolic Space
• Loss Function
• Support Vector Machine
• Hierarchical Structure
• 2D Images
• Semantic Information
• Airplane
• Root Node
• Geodesic
• Leaf Node
• Image Encoder
• Contrastive Loss
• Self-supervised Learning Methods
• Few-shot Learning
• 3D Point Cloud
• Exponential Map
• Hierarchical Information
• Classification Experiments
• Data Hierarchy
• Point Cloud Features