ROAD: Learning an Implicit Recursive Octree Auto-Decoder to Efficiently Encode 3D Shapes

12/12/2022
by   Sergey Zakharov, et al.
6

Compact and accurate representations of 3D shapes are central to many perception and robotics tasks. State-of-the-art learning-based methods can reconstruct single objects but scale poorly to large datasets. We present a novel recursive implicit representation to efficiently and accurately encode large datasets of complex 3D shapes by recursively traversing an implicit octree in latent space. Our implicit Recursive Octree Auto-Decoder (ROAD) learns a hierarchically structured latent space enabling state-of-the-art reconstruction results at a compression ratio above 99 efficient curriculum learning scheme that naturally exploits the coarse-to-fine properties of the underlying octree spatial representation. We explore the scaling law relating latent space dimension, dataset size, and reconstruction accuracy, showing that increasing the latent space dimension is enough to scale to large shape datasets. Finally, we show that our learned latent space encodes a coarse-to-fine hierarchical structure yielding reusable latents across different levels of details, and we provide qualitative evidence of generalization to novel shapes outside the training set.

READ FULL TEXT

page 2

page 15

page 16

research
09/12/2021

Multiresolution Deep Implicit Functions for 3D Shape Representation

We introduce Multiresolution Deep Implicit Functions (MDIF), a hierarchi...
research
08/10/2020

RocNet: Recursive Octree Network for Efficient 3D Deep Representation

We introduce a deep recursive octree network for the compression of 3D v...
research
09/17/2020

On the Effectiveness of Weight-Encoded Neural Implicit 3D Shapes

A neural implicit outputs a number indicating whether the given query po...
research
07/18/2022

Latent Partition Implicit with Surface Codes for 3D Representation

Deep implicit functions have shown remarkable shape modeling ability in ...
research
03/30/2020

PointGMM: a Neural GMM Network for Point Clouds

Point clouds are a popular representation for 3D shapes. However, they e...
research
02/01/2023

Neural Wavelet-domain Diffusion for 3D Shape Generation, Inversion, and Manipulation

This paper presents a new approach for 3D shape generation, inversion, a...
research
04/01/2022

LASER: LAtent SpacE Rendering for 2D Visual Localization

We present LASER, an image-based Monte Carlo Localization (MCL) framewor...

Please sign up or login with your details

Forgot password? Click here to reset