Abstract:Sampling is widely used in various point cloud tasks as it can effectively reduce resource consumption. Recently, some methods have proposed utilizing neural networks to optimize the sampling process for various task requirements. Currently, deep downsampling methods can be categorized into two main types: generative-based and score-based. Generative-based methods directly generate sampled point clouds using networks, whereas score-based methods assess the importance of points according to specific rules and then select sampled point clouds based on their scores. However, these methods often result in noticeable clustering effects in high-intensity feature areas, compromising their ability to preserve small-scale features and leading to the loss of some structures, thereby affecting the performance of subsequent tasks. In this paper, we propose REPS, a reconstruction-based scoring strategy that evaluates the importance of each vertex by removing and reconstructing them using surrounding vertices. Our reconstruction process comprises point reconstruction and shape reconstruction. The two aforementioned reconstruction methods effectively evaluate the importance of vertices by removing them at different scales for reconstruction. These reconstructions ensure that our method maintains the overall geometric features of the point cloud and avoids disturbing small-scale structures during sampling. Additionally, we propose the Global-Local Fusion Attention (GLFA) module, which aggregates local and global attention features of point clouds, ensuring high-quality reconstruction and sampling effects. Our method outperforms previous approaches in preserving the structural features of the sampled point clouds. Furthermore, abundant experimental results demonstrate the superior performance of our method across various common tasks.

Attention-based Point Cloud Edge Sampling

Attention-based Point Cloud Edge Sampling

APSNet: Attention Based Point Cloud Sampling

Fast Point Cloud Sampling Network.

PointAS: an attention based sampling neural network for visual perception

CPCS: Critical Points Guided Clustering and Sampling for Point Cloud Analysis.

REPS: Reconstruction-based Point Cloud Sampling

Learning-based Sampling Method for Point Cloud Segmentation

Rethinking Attention Module Design for Point Cloud Analysis

Attention-based Transformation from Latent Features to Point Clouds

Edge-based Segmentation of Point Cloud Data

Rethinking Sampling in 3D Point Cloud Generative Adversarial Networks

Point Attention Network for Semantic Segmentation of 3D Point Clouds

Domain Adaptive Sampling for Cross-Domain Point Cloud Recognition

Adaptive Edge-to-Edge Interaction Learning for Point Cloud Analysis

Modeling Point Clouds with Self-Attention and Gumbel Subset Sampling

Attention-Based Deep Neural Network for Point Cloud Learning

On learning the right attention point for feature enhancement

Learning Saliency Maps for Adversarial Point-Cloud Generation

Geometric attentional dynamic graph convolutional neural networks for point cloud analysis

Point Attention Network for Point Cloud Semantic Segmentation.