Pyramid Point: A Multi-Level Focusing Network for Revisiting Feature Layers
This work provides an incremental improvement in object category learning from point clouds for computer vision researchers.
The paper introduces Pyramid Point, a network designed for object category learning from unordered point sets. It utilizes a dense pyramid structure and a Focused Kernel Point convolution (FKP Conv) with an attention mechanism to improve feature quality and contextual information, achieving competitive performance on three benchmark datasets.
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.