Self-Contrastive Learning with Hard Negative Sampling for Self-supervised Point Cloud Learning

Abstract

Point clouds have attracted increasing attention. Significant progress has been made in methods for point cloud analysis, which often requires costly human annotation as supervision. To address this issue, we propose a novel self-contrastive learning for self-supervised point cloud representation learning, aiming to capture both local geometric patterns and nonlocal semantic primitives based on the nonlocal self-similarity of point clouds. The contributions are two-fold: on the one hand, instead of contrasting among different point clouds as commonly employed in contrastive learning, we exploit self-similar point cloud patches within a single point cloud as positive samples and otherwise negative ones to facilitate the task of contrastive learning. On the other hand, we actively learn hard negative samples that are close to positive samples for discriminative feature learning. Experimental results show that the proposed method achieves state-of-the-art performance on widely used benchmark datasets for self-supervised point cloud segmentation and transfer learning for classification.

arXiv commentAccepted to ACM MM 2021

References in corpus (33)

• Adam: A Method for Stochastic Optimization
• Auto-Encoding Variational Bayes
• Semi-Supervised Classification with Graph Convolutional Networks
• A Simple Framework for Contrastive Learning of Visual Representations
• Geometric deep learning: going beyond Euclidean data
• Transformers in Vision: A Survey
• Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling
• Rotation-invariant convolutional neural networks for galaxy morphology prediction
• Adversarial Feature Learning
• Learning Representations by Maximizing Mutual Information Across Views
• Adversarially Learned Inference
• Learning Representations and Generative Models for 3D Point Clouds
• MolGAN: An implicit generative model for small molecular graphs
• Learning Deep Generative Models of Graphs
• PCPNET: Learning Local Shape Properties from Raw Point Clouds
• SE(3)-Transformers: 3D Roto-Translation Equivariant Attention Networks
• Hard Negative Mixing for Contrastive Learning
• NetGAN: Generating Graphs via Random Walks
• SO-Net: Self-Organizing Network for Point Cloud Analysis
• PointContrast: Unsupervised Pre-training for 3D Point Cloud Understanding
• RGCNN: Regularized Graph CNN for Point Cloud Segmentation
• L2G Auto-encoder: Understanding Point Clouds by Local-to-Global Reconstruction with Hierarchical Self-Attention
• AET vs. AED: Unsupervised Representation Learning by Auto-Encoding Transformations rather than Data
• Contrastive Learning with Hard Negative Samples
• Self-Supervised Deep Learning on Point Clouds by Reconstructing Space
• Conditional Negative Sampling for Contrastive Learning of Visual Representations
• MortonNet: Self-Supervised Learning of Local Features in 3D Point Clouds
• Global-Local Bidirectional Reasoning for Unsupervised Representation Learning of 3D Point Clouds
• A Polynomial-time Solution for Robust Registration with Extreme Outlier Rates
• A Graph-CNN for 3D Point Cloud Classification
• SDRSAC: Semidefinite-Based Randomized Approach for Robust Point Cloud Registration without Correspondences
• Self-Supervised Multi-View Learning via Auto-Encoding 3D Transformations
• AVT: Unsupervised Learning of Transformation Equivariant Representations by Autoencoding Variational Transformations

Cited by in corpus (6)

• Unsupervised Point Cloud Representation Learning with Deep Neural Networks: A Survey
• Contrastive Learning for Joint Normal Estimation and Point Cloud Filtering
• PointSmile: Point Self-supervised Learning via Curriculum Mutual Information
• Point-DAE: Denoising Autoencoders for Self-supervised Point Cloud Learning
• ExpPoint-MAE: Better interpretability and performance for self-supervised point cloud transformers
• Self-supervised Learning of Rotation-invariant 3D Point Set Features using Transformer and its Self-distillation