English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Learning 3D generative models from a dataset of monocular images enablesself-supervised 3D reasoning and controllable synthesis. State-of-the-art 3Dgenerative models are GANs which use neural 3D volumetric representations forsynthesis. Images are synthesized by rendering the volumes from a given camera.These models can disentangle the 3D scene from the camera viewpoint in anygenerated image. However, most models do not disentangle other factors of imageformation, such as geometry and appearance. In this paper, we design a 3D GANwhich can learn a disentangled model of objects, just from monocularobservations. Our model can disentangle the geometry and appearance variationsin the scene, i.e., we can independently sample from the geometry andappearance spaces of the generative model. This is achieved using a novelnon-rigid deformable scene formulation. A 3D volume which represents an objectinstance is computed as a non-rigidly deformed canonical 3D volume. Our methodlearns the canonical volume, as well as its deformations, jointly duringtraining. This formulation also helps us improve the disentanglement betweenthe 3D scene and the camera viewpoints using a novel pose regularization lossdefined on the 3D deformation field. In addition, we further model the inversedeformations, enabling the computation of dense correspondences between imagesgenerated by our model. Finally, we design an approach to embed real imagesinto the latent space of our disentangled generative model, enabling editing ofreal images.

Disentangled3D: Learning a 3D Generative Model with Disentangled Geometry and Appearance from Monocular Images