DANCE: Density-agnostic and Class-aware Network for Point Cloud Completion

TL;DR

DANCE is a novel point cloud completion framework that is density-agnostic and class-aware, effectively recovering missing geometry while handling variable sparsity and noise without relying on external image data.

Contribution

The paper introduces DANCE, a new method that completes missing 3D structures in point clouds with density-agnostic and class-aware capabilities, improving robustness and accuracy.

Findings

Outperforms state-of-the-art methods on PCN and MVP benchmarks.

Robust to varying input densities and noise levels.

Effectively incorporates semantic guidance without external supervision.

Abstract

Point cloud completion aims to recover missing geometric structures from incomplete 3D scans, which often suffer from occlusions or limited sensor viewpoints. Existing methods typically assume fixed input/output densities or rely on image-based representations, making them less suitable for real-world scenarios with variable sparsity and limited supervision. In this paper, we introduce Density-agnostic and Class-aware Network (DANCE), a novel framework that completes only the missing regions while preserving the observed geometry. DANCE generates candidate points via ray-based sampling from multiple viewpoints. A transformer decoder then refines their positions and predicts opacity scores, which determine the validity of each point for inclusion in the final surface. To incorporate semantic guidance, a lightweight classification head is trained directly on geometric features, enabling category-consistent completion without external image supervision. Extensive experiments on the PCN and MVP benchmarks show that DANCE outperforms state-of-the-art methods in accuracy and structural consistency, while remaining robust to varying input densities and noise levels.