Research map / 3D vision / Face and shape
LSD-StructureNet: Modeling Levels of Structural Detail in 3D Part Hierarchies
Dominic Roberts, Ara Danielyan, Hang Chu, Mani Golparvar-Fard, David Forsyth
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 per-depth latent spaces (LSD) to StructureNet | StructureNet · PartNet Table conditional sampling | sampling time (s): 3.0 -> 0.04 | Improved |
| add per-depth latent spaces (LSD) to StructureNet | StructureNet · PartNet Chair conditional sampling | sampling time (s): 0.67 -> 0.05 | Improved |
| add per-depth latent spaces (LSD) to StructureNet | StructureNet · PartNet Table conditional D=1 | FPD: 85.5 -> 44.7 | Improved |
| add per-depth latent spaces (LSD) to StructureNet | StructureNet · PartNet Chair conditional D=1 | structural diversity: 73.5 -> 69.3 | Hurt |
| add per-depth latent spaces (LSD) to StructureNet | StructureNet · PartNet Storage conditional D=2 | FPD: 79.2 -> 88.1 | Hurt |
| add per-depth latent spaces (LSD) to StructureNet | StructureNet · PartNet Table unconditional (bounding boxes) | FPD: 103.2 -> 36.3 | Improved |
| add per-depth latent spaces (LSD) to StructureNet | StructureNet · PartNet Chair unconditional (bounding boxes) | FPD: 57.0 -> 39.3 | Improved |
| model all depths probabilistically instead of only depth <=1 (ablation) | LSD-StructureNet · PartNet Chair | quality: 76.5 -> 47.2 | Improved |
| model all depths probabilistically instead of only depth <=3 (ablation) | LSD-StructureNet · PartNet Chair | coverage: 32.1 -> 25.4 | Improved |
| add encoder and decoder LSTMs (ablation) | LSD-StructureNet · PartNet Chair | coverage: 73.5 -> 25.4 | Improved |
| add per-depth latent spaces (LSD) to StructureNet | StructureNet · PartNet Chair vs PQ-Net | quality: 116.5 -> 47.2 | Improved |
| use LSD-StructureNet instead of PQ-Net | PQ-Net · PartNet Chair | FPD: 28.9 -> 39.3 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- PQ-Net — PQ-NET: A Generative Part Seq2Seq Network for 3D Shapes (CVPR 2020) · better on 0 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Generative 3D Part Assembly via Part-Whole-Hierarchy Message Passing (CVPR 2024 · similarity 0.89)
- PartGen: Part-level 3D Generation and Reconstruction with Multi-view Diffusion Models (CVPR 2025 · similarity 0.88)
- PartCrafter: Structured 3D Mesh Generation via Compositional Latent Diffusion Transformers (NeurIPS 2025 · similarity 0.88)
- PartNeXt: A Next-Generation Dataset for Fine-Grained and Hierarchical 3D Part Understanding (NeurIPS 2025 · similarity 0.88)
- Learning Partonomic 3D Reconstruction from Image Collections (CVPR 2025 · similarity 0.87)
- 3DCoMPaT200: Language Grounded Large-Scale 3D Vision Dataset for Compositional Recognition (NeurIPS 2024 · similarity 0.87)
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.
