Research map / Detection and segmentation / Open-set and few-shot detection
LEGO-Net: Learning Regular Rearrangements of Objects in Rooms
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| replace one-pass direct regression with iterative denoising (ablation) | LEGO-Net · Table-Chair: Symmetry & Parallelism | success rate: 16% -> 91.2% | Helped |
| replace one-pass direct regression with iterative denoising (ablation) | LEGO-Net · Table-Chair: Uniform Spacing | success rate: 18.6% -> 96% | Helped |
| replace one-pass direct regression with iterative denoising (ablation) | LEGO-Net · Table-Chair: Grouping by Shape | success rate: 23.6% -> 87.8% | Helped |
| add Langevin noise during iterative denoising (ablation) | LEGO-Net · Table-Chair: Symmetry & Parallelism; Uniform Spacing; Grouping by Shape | success rate: 91.2% -> 91.4%; 96% -> 97.2%; 87.8% -> 89.2% | Helped |
| omit Langevin noise during iterative denoising (ablation) | LEGO-Net · 3D-FRONT Living Room and Bedroom | KID: 67 -> 51; 37 -> 27 | Helped |
| omit Langevin noise during iterative denoising (ablation) | LEGO-Net · 3D-FRONT Living Room and Bedroom | EMD to GT: 0.125 -> 0.117; 0.086 -> 0.0815 | Helped |
| replace PointNet with ResNet for floor-plan encoding (ablation) | LEGO-Net · 3D-FRONT floor-boundary placement | percentage of scenes with 90% of furniture within boundaries: 84.4% -> 83.8%; 54.4% -> 54% | Hurt |
| add noise to the PointNet floor-plan encoder (ablation) | LEGO-Net · 3D-FRONT floor-boundary placement | percentage of scenes with 90% of furniture within boundaries: 84.4% -> 84.2%; 54.4% -> 54.2% | Hurt |
| predict relative rather than absolute transformations (ablation) | LEGO-Net · Table-Chair: Uniform & Parallelism | Distance Moved: 2.98e-1 -> 2.47e-1 | Helped |
| predict relative rather than absolute transformations (ablation) | LEGO-Net · Table-Chair: Uniform & Parallelism | EMD to GT: 5.16e-2 -> 1.58e-1 | Hurt |
| use LEGO-Net iterative denoising rather than ATISS failure-correction | Room rearrangement · 3D-FRONT Living Room and Bedroom | KID: 280 -> 51; 240 -> 27 | Helped |
| use LEGO-Net iterative denoising rather than ATISS failure-correction | Room rearrangement · 3D-FRONT Living Room and Bedroom | Distance Moved: 0.1473 -> 0.086; 0.2025 -> 0.0492 | Helped |
Closest papers by meaning
- DiffuScene: Denoising Diffusion Models for Generative Indoor Scene Synthesis (CVPR 2024 · similarity 0.87)
- A Category Agnostic Model for Visual Rearrangment (CVPR 2024 · similarity 0.87)
- OptiScene: LLM-driven Indoor Scene Layout Generation via Scaled Human-aligned Data Synthesis and Multi-Stage Preference Optimization (NeurIPS 2025 · similarity 0.86)
- DeBaRA: Denoising-Based 3D Room Arrangement Generation (NeurIPS 2024 · similarity 0.86)
- Disentangled 3D Scene Generation with Layout Learning (ICML 2024 · similarity 0.86)
- Task Planning for Object Rearrangement in Multi-Room Environments (AAAI 2024 · 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.