Understanding about 3D convolution and pool process on octree about o-cnn HOT 2 CLOSED

O-CNN constrains the CNN storage and computation into non-empty sparse voxels for efficiency and uses the octree data structure to organize and index these sparse voxels. Currently, this type of 3D convolution is known as Sparse Convolution in the research community.
The concept of Sparse Convolution is almost the same in our O-CNN in SIGGRAPH 2017, SparseConvNet in CVPR 2018, and MinkowskiNet in CVPR 2019, except that we use Octrees to index sparse non-empty voxel while SparseConvNet and MinowskiNet use Hash Tables.

Here is a ppt containing animations about O-CNN. Hope it may be helpful to you.

from o-cnn.

O-CNN constrains the CNN storage and computation into non-empty sparse voxels for efficiency and uses the octree data structure to organize and index these sparse voxels. Currently, this type of 3D convolution is known as Sparse Convolution in the research community. The concept of Sparse Convolution is almost the same in our O-CNN in SIGGRAPH 2017, SparseConvNet in CVPR 2018, and MinkowskiNet in CVPR 2019, except that we use Octrees to index sparse non-empty voxel while SparseConvNet and MinowskiNet use Hash Tables.

Here is a ppt containing animations about O-CNN. Hope it may be helpful to you.

Thank you for your guidance, I will dive into these references.

from o-cnn.