跳到正文
arXiv cs.AI· Jiawei Shen, Jie Feng, Junpeng Zhang, Guanbin Li, Wanyi Chen, Weisheng Dong, Guangming Shi·· 1 小时前AI 评分29

ReflectScene:MLLM 引导的闭环扩散框架用于 3D 室内场景生成

ReflectScene:MLLM-Guided Closed-Loop Diffusion for 3D Indoor Scene Generation

AI 导读

ReflectScene 是一个由多模态大语言模型(MLLM)引导的闭环扩散框架,用于文本条件下的 3D 室内场景生成。它通过分层先验场景理解、MLLM 引导的语义奖励评估以及决策条件化的轨迹一致扩散,在去噪过程中修正物体布局并保持场景一致性。在 3D-FRONT 数据集上的实验显示,其功能一致性与语义对齐均有提升,同时保持有竞争力的生成质量与多样性。

正文 · 原文

View PDF HTML (experimental)

Abstract:Text-conditioned 3D indoor scene generation requires layouts satisfying geometric, functional, and semantic constraints. However, existing diffusion methods rely primarily on static textual conditions, limiting their ability to assess evolving scenes and translate semantic discrepancies into executable object-level corrections. MLLM-based post-hoc refinement provides semantic feedback but remains decoupled from denoising, potentially disrupting scene coherence. We propose ReflectScene, an MLLM-guided closed-loop diffusion framework. First, we introduce Hierarchical Prior-Grounded Scene Understanding that aligns structured layouts with multi-view visual evidence, to interpret object-specific context and scene global semantics via a frozen multi-modal large language model (MLLM). Second, we propose MLLM-Guided Semantic Reward Evaluation to distill MLLM's diagnoses and assessments on possible candidate corrections into a lightweight Reward Predictor, enabling correction decision making with reduced computation burden. Finally, Decision-Conditioned Trajectory-Consistent Diffusion is proposed to integrate selected corrections through transient conditioning and persistent clean-estimate projection, correcting object layout in the generated scene while preserving non-target arrangements throughout denoising. Experiments on 3D-FRONT demonstrate improved functional consistency and semantic alignment while maintaining competitive generation quality and diversity.
Comments: 10 pages, 3 figures
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Cite as: arXiv:2611.23845 [cs.CV]
  (or arXiv:2611.23845v2 [cs.CV] for this version)
  https://doi.org/10.48550/arXiv.2611.23845

arXiv-issued DOI via DataCite

Submission history

From: Jiawei Shen [view email]
[v1] Wed, 18 Nov 2026 04:50:50 UTC (17,114 KB)
[v2] Thu, 19 Nov 2026 05:59:35 UTC (17,114 KB)

来源:arXiv cs.AI · arxiv.org