DinoComplete: 3D Shape Completion with Distilled Semantic Priors and State Space Models

2026-05-26

AI Summary

This work addresses the challenging problem of 3D shape completion under unseen categories and real-world noisy scans by proposing an efficient deterministic method that integrates geometric and semantic priors. It introduces a novel approach to distill semantic knowledge from the DINO vision foundation model into voxel-aligned dense features and designs a multi-scale Voxel Mamba module to enable long-range contextual modeling at high resolution. Through a unified geometry–semantics fusion architecture, the method significantly outperforms existing deterministic and generative approaches on both ShapeNet with unseen categories and real-world ScanNet data. Notably, it achieves superior generalization and robustness while using fewer parameters, lower memory consumption, and faster inference speed.

Abstract

3D shape completion from partial scans remains challenging for unseen categories and noisy real-world observations, where geometry alone is often insufficient for inferring missing structure. We present DinoComplete, a deterministic and efficient shape completion framework that augments geometric reconstruction with voxel-aligned semantic priors distilled from DINO features. First, we construct multi-view DINO feature volumes aligned with ShapeNet data and train a student network to predict dense semantic features directly from incomplete shapes. These predicted features capture global structure and part-aware semantic context while remaining aligned with the underlying geometry. We then integrate these distilled features into a completion network, where geometric and semantic voxel representations are fused through voxel state-space modeling. To enable efficient long-range reasoning without sacrificing resolution, we introduce a multi-scale voxel Mamba module that refines the fused features by combining full-grid and chunk-wise sequence modeling. Experiments on unseen ShapeNet categories and ScanNet objects show that DinoComplete achieves stronger completion quality than prior deterministic and generative based completion methods while using fewer parameters, requiring lower memory, and achieving faster inference. Our results demonstrate that distilling semantic priors from visual foundation models improves generalization and robustness in 3D shape completion.

Problem

• 3D shape completion
• partial scans
• unseen categories
• noisy observations
• semantic priors

Innovation

• semantic priors distillation
• voxel state-space modeling
• multi-scale Mamba
• 3D shape completion
• DINO features