SoftPoolNet: Shape Descriptor for Point Cloud Completion and Classification

TL;DR

This paper introduces SoftPoolNet, a novel point cloud processing method that enhances 3D object completion and classification by using soft pooling, regional convolutions, and patch-deforming operations, achieving state-of-the-art results.

Contribution

The paper presents a new feature organization method called soft pooling, regional convolutions for maximizing activation entropy, and a patch-deforming operation for improved point cloud processing.

Findings

Achieves state-of-the-art accuracy on 3D object completion and classification tasks.

Incorporating regional activation improves performance across architectures.

Proves the effectiveness of the proposed methods on multiple 3D benchmarks.

Abstract

Point clouds are often the default choice for many applications as they exhibit more flexibility and efficiency than volumetric data. Nevertheless, their unorganized nature -- points are stored in an unordered way -- makes them less suited to be processed by deep learning pipelines. In this paper, we propose a method for 3D object completion and classification based on point clouds. We introduce a new way of organizing the extracted features based on their activations, which we name soft pooling. For the decoder stage, we propose regional convolutions, a novel operator aimed at maximizing the global activation entropy. Furthermore, inspired by the local refining procedure in Point Completion Network (PCN), we also propose a patch-deforming operation to simulate deconvolutional operations for point clouds. This paper proves that our regional activation can be incorporated in many point cloud architectures like AtlasNet and PCN, leading to better performance for geometric completion. We evaluate our approach on different 3D tasks such as object completion and classification, achieving state-of-the-art accuracy.