Deep Deformable Models: Learning 3D Shape Abstractions with Part Consistency

TL;DR

This paper introduces Deep Deformable Models (DDMs), a novel approach for 3D shape abstraction that uses flexible primitives with part consistency, enabling more accurate and semantically meaningful shape representations.

Contribution

The paper proposes DDM, which employs global and local deformations for shape abstraction, offering improved geometric flexibility and part-level semantic correspondence over existing methods.

Findings

DDM outperforms state-of-the-art in shape reconstruction accuracy.

DDM achieves better part consistency in shape abstractions.

Extensive experiments on ShapeNet validate the effectiveness of DDM.

Abstract

The task of shape abstraction with semantic part consistency is challenging due to the complex geometries of natural objects. Recent methods learn to represent an object shape using a set of simple primitives to fit the target. \textcolor{black}{However, in these methods, the primitives used do not always correspond to real parts or lack geometric flexibility for semantic interpretation.} In this paper, we investigate salient and efficient primitive descriptors for accurate shape abstractions, and propose \textit{Deep Deformable Models (DDMs)}. DDM employs global deformations and diffeomorphic local deformations. These properties enable DDM to abstract complex object shapes with significantly fewer primitives that offer broader geometry coverage and finer details. DDM is also capable of learning part-level semantic correspondences due to the differentiable and invertible properties of our primitive deformation. Moreover, DDM learning formulation is based on dynamic and kinematic modeling, which enables joint regularization of each sub-transformation during primitive fitting. Extensive experiments on \textit{ShapeNet} demonstrate that DDM outperforms the state-of-the-art in terms of reconstruction and part consistency by a notable margin.