3D Indoor Scene Synthesis on Dining + Study Average (Generalization Split)
17OBR (Object Realism)I-Design
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| I-DesignMethod Category=LLM with post-processing2026.02 | 17 | 2.6 | 8 | 6.3 | 8.9 | 7.7 | |
| LayoutVLMMethod Category=LLM with post-processing2026.02 | 15.8 | 31.8 | 6.7 | 7.7 | 8.7 | 7.7 | |
| HolodeckMethod Category=LLM with post-processing2026.02 | 12.8 | 4.2 | 6.6 | 8.2 | 8.7 | 7.8 | |
| LayoutGPTMethod Category=LLM-based few-shot learners2026.02 | 11.4 | 39.2 | 7.7 | 7.5 | 8.6 | 7.9 | |
| SceneReVisPolicy=Fine-tuned Qwen2.5-VL-7B, Reward Model=Qwen3-VL-235B, Algorithm=GRPO2026.02 | 0.3 | 2.5 | 8.2 | 8.3 | 8.8 | 8.4 |