Research map / 3D vision / Face and shape
Unified Primitive Proxies for Structured Shape Completion
Zhaiyu Chen, Yuqing Wang, Xiao Xiang Zhu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add online target updates (ablation) | UniCo · ABC-multi ablation (200 epochs) | CD x100: 12.22 -> 2.44 | Improved |
| add online target updates (ablation) | UniCo · ABC-multi ablation (200 epochs) | NC: 0.631 -> 0.924 | Improved |
| train both pathways jointly instead of two-stage (ablation) | UniCo · ABC-multi ablation (200 epochs) | CD x100: 2.55 -> 2.44 | Improved |
| add parameter head (ablation) | UniCo · ABC-multi ablation (200 epochs) | CD x100: 2.52 -> 2.44 | Improved |
| add primitive-wise Chamfer (ablation) | UniCo · ABC-multi ablation (200 epochs) | CD x100: 2.53 -> 2.44 | Improved |
| use cross-entropy plus Dice for membership instead of CE only (ablation) | UniCo · ABC-multi ablation (200 epochs) | CD x100: 2.53 -> 2.44 | Improved |
| use cross-entropy plus Dice for membership instead of Dice only (ablation) | UniCo · ABC-multi ablation (200 epochs) | NC: 0.914 -> 0.924 | Improved |
| use 40 proxies instead of 30 (ablation) | UniCo · ABC-multi proxy count | CD x100: 2.47 -> 2.44 | Improved |
| use no-object weight 0.01 instead of 0.05 (ablation) | UniCo · ABC-multi proxy count | CD x100: 2.73 -> 2.44 | Improved |
| confidence threshold 0.5 instead of 0.3 or 0.7 (ablation) | UniCo · ABC-multi confidence threshold | CD x100: 2.48 -> 2.44 | No clear effect |
| add projection post-processing | UniCo · ABC-plane (PolyFit) | CD x100: 1.69 -> 1.67 | No clear effect |
| replace cascaded PaCo with joint pathways | UniCo · Building-PCC (PolyFit) | CD x100: 4.89 -> 3.84 | Improved |
Closest papers by meaning
- Unsupervised 3D Shape Reconstruction by Part Retrieval and Assembly (CVPR 2023 · similarity 0.87)
- PatchRD: Detail-Preserving Shape Completion by Learning Patch Retrieval and Deformation (ECCV 2022 · similarity 0.87)
- ME-PCN: Point Completion Conditioned on Mask Emptiness (ICCV 2021 · similarity 0.87)
- Neural Parts: Learning Expressive 3D Shape Abstractions With Invertible Neural Networks (CVPR 2021 · similarity 0.87)
- DinoComplete: 3D Shape Completion with Distilled Semantic Priors and State Space Models (NeurIPS 2026 · similarity 0.85)
- ESCAPE: Equivariant Shape Completion via Anchor Point Encoding (CVPR 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
