【点云论文速读】RandLA-Net:大场景三维点云语义分割新框架

标题:RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds

作者:Qingyong Hu, Bo Yang , Linhai Xie, Stefano Rosa, Yulan Guo

来源:2020 CVPR Oral

星球ID:大志_ETU_分割

欢迎各位加入免费知识星球,获取PDF论文,欢迎转发朋友圈分享快乐。

论文摘要

这篇论文提出了一种基于随机降采样和局部特征聚合的网络结构(RandLA-Net)。并在Semantic3D和SemanticKITTI等大场景点云分割数据集上取得了较好的效果,具有很高的效率(作者实验称相比基于图的方法SPG效率提升200倍)。

主要贡献:

1. 分析和比较了现有的点云降采样方法,将降采样方式分为Heuristic Sampling以及Learning-based Sampling两大类。并且认为随机降采样是一种适合大规模点云高效学习的方法(目前来看是)。

2. 为了降低随机采样的信息丢失,进一步提出了局部特征聚合模块(Local feature aggregation),包括三个子模块:

1) 局部空间编码(LocSE):显式地编码三维点云的空间几何形状信息,网络能够从各个点的相对位置以及距离信息中更好地学习到空间的几何结构;

2) 注意力池化(attentive pooling):上一步的邻域特征点集的加权求和,进行特征聚合(特征深度融合);

3) 扩张残差块(dilated residual block): 增大每个点的感受野,通过逐步增加每个点的感受野来更好地学习和保留大场景点云中复杂的几何结构。

RandLA-Net网络在多个大场景点云的数据集上都展现出较好的效果,及较高的计算效率,已开源。

论文图集

英文摘要

We study the problem of efficient semantic segmentation for large-scale 3D point clouds. By relying on expensive sampling techniques or computationally heavy pre/postprocessing steps, most existing approaches are only able to be trained and operate over small-scale point clouds.
In this paper, we introduce RandLA-Net, an efficient and lightweight neural architecture to directly infer per-point semantics for large-scale point clouds. The key to our approach is to use random point sampling instead of more complex point selection approaches. Although remarkably computation and memory efficient, random sampling can discard key features by chance. To overcome this, we introduce a novel local feature aggregation module to progressively increase the receptive field for each 3D point, thereby effectively preserving geometric details. Extensive experiments show that our RandLA-Net can process 1 million points in a single pass with up to 200× faster than existing approaches. Moreover, our RandLA-Net clearly surpasses state-of-the-art approaches for semantic segmentation on two large-scale benchmarks Semantic3D and SemanticKITTI.

(0)

相关推荐