OmniShape: Zero-Shot Multi-Hypothesis Shape and Pose Estimation in the Real World

TL;DR

OmniShape introduces a novel zero-shot approach for probabilistic shape and pose estimation from a single image, leveraging separate diffusion models for shape and pose hypotheses without requiring known models or categories.

Contribution

It is the first method to enable joint probabilistic shape and pose estimation in a zero-shot setting using decoupled multi-modal distributions.

Findings

Effective on real-world datasets

Samples multiple hypotheses for shape and pose

Outperforms existing methods in zero-shot scenarios

Abstract

We would like to estimate the pose and full shape of an object from a single observation, without assuming known 3D model or category. In this work, we propose OmniShape, the first method of its kind to enable probabilistic pose and shape estimation. OmniShape is based on the key insight that shape completion can be decoupled into two multi-modal distributions: one capturing how measurements project into a normalized object reference frame defined by the dataset and the other modelling a prior over object geometries represented as triplanar neural fields. By training separate conditional diffusion models for these two distributions, we enable sampling multiple hypotheses from the joint pose and shape distribution. OmniShape demonstrates compelling performance on challenging real world datasets. Project website: https://tri-ml.github.io/omnishape