DV-ConvNet: Fully Convolutional Deep Learning on Point Clouds with Dynamic Voxelization and 3D Group Convolution
arXiv:2009.02918
Abstract
3D point cloud interpretation is a challenging task due to the randomness and sparsity of the component points. Many of the recently proposed methods like PointNet and PointCNN have been focusing on learning shape descriptions from point coordinates as point-wise input features, which usually involves complicated network architectures. In this work, we draw attention back to the standard 3D convolutions towards an efficient 3D point cloud interpretation. Instead of converting the entire point cloud into voxel representations like the other volumetric methods, we voxelize the sub-portions of the point cloud only at necessary locations within each convolution layer on-the-fly, using our dynamic voxelization operation with self-adaptive voxelization resolution. In addition, we incorporate 3D group convolution into our dense convolution kernel implementation to further exploit the rotation invariant features of point cloud. Benefiting from its simple fully-convolutional architecture, our network is able to run and converge at a considerably fast speed, while yields on-par or even better performance compared with the state-of-the-art methods on several benchmark datasets.
References in corpus (8)
- Adam: A Method for Stochastic Optimization
- Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point clouds
- Submanifold Sparse Convolutional Networks
- FPNN: Field Probing Neural Networks for 3D Data
- PointConv: Deep Convolutional Networks on 3D Point Clouds
- N-body Networks: a Covariant Hierarchical Neural Network Architecture for Learning Atomic Potentials
- Deep Learning for 3D Point Clouds: A Survey
- Octree guided CNN with Spherical Kernels for 3D Point Clouds