Unsupervised Kinematic Motion Detection for Part-segmented 3D Shape Collections

TL;DR

This paper introduces an unsupervised method to discover articulated part motions in 3D shape collections by leveraging category consistency, outperforming some supervised methods on the PartNet-Mobility dataset.

Contribution

It proposes a novel unsupervised algorithm based on category closure to predict part motions in 3D models without requiring annotated training data.

Findings

Achieves low error in predicted motions compared to ground truth.

Outperforms two supervised motion prediction methods.

Works effectively across multiple shape categories.

Abstract

3D models of manufactured objects are important for populating virtual worlds and for synthetic data generation for vision and robotics. To be most useful, such objects should be articulated: their parts should move when interacted with. While articulated object datasets exist, creating them is labor-intensive. Learning-based prediction of part motions can help, but all existing methods require annotated training data. In this paper, we present an unsupervised approach for discovering articulated motions in a part-segmented 3D shape collection. Our approach is based on a concept we call category closure: any valid articulation of an object's parts should keep the object in the same semantic category (e.g. a chair stays a chair). We operationalize this concept with an algorithm that optimizes a shape's part motion parameters such that it can transform into other shapes in the collection. We evaluate our approach by using it to re-discover part motions from the PartNet-Mobility dataset. For almost all shape categories, our method's predicted motion parameters have low error with respect to ground truth annotations, outperforming two supervised motion prediction methods.