Pyramid Point: A Multi-Level Focusing Network for Revisiting Feature Layers

TL;DR

Pyramid Point introduces a multi-level focusing network with a dense pyramid structure and attention-based convolutions, enhancing feature learning for object classification from point clouds and achieving competitive results.

Contribution

The paper proposes a novel dense pyramid architecture and Focused Kernel Point convolution with attention, improving feature quality and contextual understanding in point cloud segmentation.

Findings

Achieves competitive performance on benchmark datasets.

Demonstrates the effectiveness of the FKP Conv and pyramid structure through ablation studies.

Enhances feature representation by revisiting layers simultaneously.

Abstract

We present a method to learn a diverse group of object categories from an unordered point set. We propose our Pyramid Point network, which uses a dense pyramid structure instead of the traditional 'U' shape, typically seen in semantic segmentation networks. This pyramid structure gives a second look, allowing the network to revisit different layers simultaneously, increasing the contextual information by creating additional layers with less noise. We introduce a Focused Kernel Point convolution (FKP Conv), which expands on the traditional point convolutions by adding an attention mechanism to the kernel outputs. This FKP Conv increases our feature quality and allows us to weigh the kernel outputs dynamically. These FKP Convs are the central part of our Recurrent FKP Bottleneck block, which makes up the backbone of our encoder. With this distinct network, we demonstrate competitive performance on three benchmark data sets. We also perform an ablation study to show the positive effects of each element in our FKP Conv.