SHRED: 3D Shape Region Decomposition with Learned Local Operations

TL;DR

SHRED is a novel method for 3D shape region decomposition that uses learned local operations to produce high-quality, fine-grained segmentations from point clouds, generalizing well to unseen categories.

Contribution

SHRED introduces a set of learned local operations for 3D shape segmentation, enabling category-agnostic, fine-grained part decomposition with improved accuracy over baselines.

Findings

SHRED outperforms baseline methods in segmentation quality.

SHRED generalizes to unseen categories without retraining.

SHRED enhances downstream segmentation tasks in zero-shot and few-shot settings.

Abstract

We present SHRED, a method for 3D SHape REgion Decomposition. SHRED takes a 3D point cloud as input and uses learned local operations to produce a segmentation that approximates fine-grained part instances. We endow SHRED with three decomposition operations: splitting regions, fixing the boundaries between regions, and merging regions together. Modules are trained independently and locally, allowing SHRED to generate high-quality segmentations for categories not seen during training. We train and evaluate SHRED with fine-grained segmentations from PartNet; using its merge-threshold hyperparameter, we show that SHRED produces segmentations that better respect ground-truth annotations compared with baseline methods, at any desired decomposition granularity. Finally, we demonstrate that SHRED is useful for downstream applications, out-performing all baselines on zero-shot fine-grained part instance segmentation and few-shot fine-grained semantic segmentation when combined with methods that learn to label shape regions.