Latent feature disentanglement for 3D meshes
This work addresses the need for better 3D shape generation and manipulation in fields like Computer Vision and Graphics, but it is incremental as it builds upon existing 3D mesh-convolutional Variational AutoEncoders.
The paper tackled the problem of generative modeling for 3D meshes by introducing a supervised model that disentangles latent shape representations into independent factors, resulting in improved random shape generation and successful application to tasks like pose and shape transfer.
Generative modeling of 3D shapes has become an important problem due to its relevance to many applications across Computer Vision, Graphics, and VR. In this paper we build upon recently introduced 3D mesh-convolutional Variational AutoEncoders which have shown great promise for learning rich representations of deformable 3D shapes. We introduce a supervised generative 3D mesh model that disentangles the latent shape representation into independent generative factors. Our extensive experimental analysis shows that learning an explicitly disentangled representation can both improve random shape generation as well as successfully address downstream tasks such as pose and shape transfer, shape-invariant temporal synchronization, and pose-invariant shape matching.