Dual Octree Graph Networks for Learning Adaptive Volumetric Shape Representations
Event Type
Technical Paper
Interest Areas
Research & Education
Presentation Types
Registration Categories
Full Conference Supporter
Full Conference
Virtual Conference Supporter
Virtual Conference
Exhibitor Additional Full Conference
Exhibitor Full Conference
DescriptionGiven an input point cloud, our network can predict the reconstructed mesh by just one forward pass and with state-of-the-art performances. To this end, we use an octree to partition the 3D volume and build a novel graph network on the dual octree to learn the adaptive volumetric shape representation.