Research map / 3D vision / 3D generation
ATISS: Autoregressive Transformers for Indoor Scene Synthesis
Despoina Paschalidou, Amlan Kar, Maria Shugrina, Karsten Kreis, Andreas Geiger, Sanja Fidler
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 |
|---|---|---|---|
| Use unordered, permutation-invariant generation rather than fixed frequency ordering (ablation) | ATISS · Bedrooms | Scene classification accuracy: 0.7603 -> 0.5620 | Improved |
| Train with random permutations and positional embeddings instead of the permutation-invariant model (ablation) | ATISS · Bedrooms | FID: 38.39 -> 40.18 | Hurt |
| Use one logistic component instead of K=10 (ablation) | ATISS · Bedrooms | FID: 38.39 -> 41.71 | Hurt |
| Increase logistic components from K=10 to K=15 (ablation) | ATISS · Bedrooms | FID: 38.39 -> 40.41 | Hurt |
| Increase logistic components from K=10 to K=20 (ablation) | ATISS · Bedrooms | Scene classification accuracy: 0.5620 -> 0.6680 | Hurt |
| Replace the ResNet-18 layout encoder with AlexNet (ablation) | ATISS · Bedrooms | FID: 38.39 -> 40.40 | Hurt |
| Use ATISS instead of Ours+Order | ATISS · Living | FID: 35.37 -> 33.14 | Improved |
| Use ATISS instead of Ours+Order | ATISS · Dining | FID: 35.79 -> 29.23 | Improved |
| Use ATISS instead of Ours+Order | ATISS · Library | Scene classification accuracy: 0.572 -> 0.521 | Improved |
| Use ATISS instead of SceneFormer | ATISS · Bedrooms | Category KL divergence: 0.0052 -> 0.0085 | Hurt |
Later papers that compared with this method
13 later accepted papers report ATISS as a baseline in their results tables; 13 of them report a better number on most of the results they share with it.
- FurniScene: A Large-scale 3D Room Dataset with Intricate Furnishing Scenes (IJCV 2026 · better on 7 of 7 shared results)
- SPREAD: Spatial-Physical REasoning via geometry Aware Diffusion (CVPR 2026 · better on 6 of 6 shared results)
- Pair2Scene: Learning Local Object Relations for Procedural Scene Generation (ICML 2026 · better on 2 of 2 shared results)
- CASAGPT: Cuboid Arrangement and Scene Assembly for Interior Design (CVPR 2025 · better on 9 of 9 shared results)
- InstructLayout: Instruction-Driven 2D and 3D Layout Synthesis With Semantic Graph Prior (TPAMI 2025 · better on 6 of 6 shared results)
- From Programs to Poses: Factored Real-World Scene Generation via Learned Program Libraries (NeurIPS 2025 · better on 4 of 4 shared results)
- Hierarchically-Structured Open-Vocabulary Indoor Scene Synthesis with Pre-trained Large Language Model (AAAI 2025 · better on 2 of 2 shared results)
- DiffuScene: Denoising Diffusion Models for Generative Indoor Scene Synthesis (CVPR 2024 · better on 9 of 9 shared results)
5 more not listed.
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
- DeBaRA: Denoising-Based 3D Room Arrangement Generation (NeurIPS 2024 · similarity 0.90)
- OptiScene: LLM-driven Indoor Scene Layout Generation via Scaled Human-aligned Data Synthesis and Multi-Stage Preference Optimization (NeurIPS 2025 · similarity 0.90)
- Direct Numerical Layout Generation for 3D Indoor Scene Synthesis via Spatial Reasoning (NeurIPS 2025 · similarity 0.90)
- CASAGPT: Cuboid Arrangement and Scene Assembly for Interior Design (CVPR 2025 · similarity 0.89)
- S-INF: Towards Realistic Indoor Scene Synthesis via Scene Implicit Neural Field (AAAI 2025 · similarity 0.89)
- SceneCraft: Layout-Guided 3D Scene Generation (NeurIPS 2024 · similarity 0.88)
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.
