3D SceneFlowNet: Self-Supervised 3D Scene Flow Estimation Based on Graph CNN
3D SceneFlowNet: Self-Supervised 3D Scene Flow Estimation Based on Graph CNN
复制标题
DOI：
10.1109/icip42928.2021.9506286
复制
发表时间：
2021-09
期刊：
影响因子：
--
通讯作者：
Yawen Lu;Yuhao Zhu;G. Lu
中科院分区：
文献类型：
--
作者：
Yawen Lu;Yuhao Zhu;G. Lu
尽管深度学习方法在2D光流估计方面取得了令人鼓舞的成功，但由于点云本身缺乏拓扑信息，因此准确估计3D空间中的场景流是一项挑战。在本文中，我们的目标是处理基于动态图卷积神经网络（GCNN）的自监督3D场景流估计问题，即3D SceneFlowNet。为了更好地学习点之间的几何关系，我们引入了EdgeConv来从点云中学习金字塔中的多层特征，并引入了自注意机制来应用多层特征来预测最终的场景流。我们的训练模型可以有效地处理一对相邻的点云作为输入，并在没有任何监督的情况下准确地预测3D场景流。所提出的方法在合成ModelNet40数据集和来自KITTI Scene Flow 2015数据集的真实的LiDAR扫描上都实现了上级性能。
Despite deep learning approaches have achieved promising successes in 2D optical flow estimation, it is a challenge to accurately estimate scene flow in 3D space as point clouds are inherently lacking topological information. In this paper, we aim at handling the problem of self-supervised 3D scene flow estimation based on dynamic graph convolutional neural networks (GCNNs), namely 3D SceneFlowNet. To better learn geometric relationships among points, we introduce EdgeConv to learn multiple-level features in a pyramid from point clouds and a self-attention mechanism to apply the multi-level features to predict the final scene flow. Our trained model can efficiently process a pair of adjacent point clouds as input and predict a 3D scene flow accurately without any supervision. The proposed approach achieves superior performance on both synthetic ModelNet40 dataset and real LiDAR scans from KITTI Scene Flow 2015 datasets.