pointflow: 3d point cloud generation with continuous normalizing flows

The videos of the transformations can be viewed on our project website: https:// www guandaoyang com/ PointFlow/ Abstract As 3D point clouds become the 

PointFlow: 3D Point Cloud Generation with Continuous Normalizing Flows Guandao Yang, Xun Huang, Zekun Hao, Ming-Yu Liu, Serge Belongie, Bharath  

malizing flows with affine coupling layers to generate 3D point clouds of an arbitrary size flow, which are referred to as “continuous normalizing flows” Several 

we also propose SoftPointFlow for 3D point cloud generation which relieves the to the existing flow-based models including both discrete and continuous normalizing flows Figure 7: Examples of point clouds generated by PointFlow and 

C-Flow: Conditional Generative Flow Models for Images and 3D Point Clouds novel conditioning scheme that brings normalizing flows to an entirely log- likelihood to model a continuous latent variable with convolutions for image generation and editing however, cannot condition the shape, preventing PointFlow

nuity, are robust to variable and irregularly spacetime-sampled point clouds, and shapes using continuous normalizing flows encode spatiotemporal (ST) changes in object shape as seen from a 3D sensor This is varying camera pose or temporal dynamics, and support shape generation at arbitrary spatiotemporal

Reasoning over an instance composed of a set of vectors, like a point cloud, requires Pointflow: 3d point cloud generation with continuous normalizing flows

erating 3D point clouds that leverage properties of 3D point cloud generation, and we believe it opens a new PointFlow uses continuous normalizing flow

erating 3D point clouds that leverage properties of successes, generating 3D point clouds with deep learning PointFlow uses continuous normalizing flow