Unified Primitive Proxies for Structured Shape Completion

TL;DR

UniCo introduces a unified primitive-based approach for structured shape completion, predicting complete primitives with semantics and inlier info in a single pass, significantly improving accuracy over recent methods.

Contribution

The paper proposes a novel unified primitive proxy framework that predicts complete shape primitives in a single forward pass, enhancing structured shape completion performance.

Findings

Reduces Chamfer distance by up to 50%

Improves normal consistency by up to 7%

Outperforms recent baselines on synthetic and real-world benchmarks

Abstract

Structured shape completion recovers missing geometry as primitives rather than as unstructured points, which enables primitive-based surface reconstruction. Instead of following the prevailing cascade, we rethink how primitives and points should interact, and find it more effective to decode primitives in a dedicated pathway that attends to shared shape features. Following this principle, we present UniCo, which in a single feed-forward pass predicts a set of primitives with complete geometry, semantics, and inlier membership. To drive this unified representation, we introduce primitive proxies, learnable queries that are contextualized to produce assembly-ready outputs. To ensure consistent optimization, our training strategy couples primitives and points with online target updates. Across synthetic and real-world benchmarks with four independent assembly solvers, UniCo consistently outperforms recent baselines, lowering Chamfer distance by up to 50% and improving normal consistency by up to 7%. These results establish an attractive recipe for structured 3D understanding from incomplete data. Project page: https://unico-completion.github.io.