用场景图指导图像生成，提升布局准确性与真实感。
SATURN: Autoregressive Image Generation Guided by Scene Graphs
- 将场景图转为重要性排序的令牌序列，引导自回归模型生成
- 在Visual Genome上FID降至21.62，得分提升至24.78
- 轻量级设计，仅微调变换器，适合追求高效高质生成的场景
当前顶尖文本到图像模型在生成逼真图像方面表现优异，但常难以捕捉复杂提示中的布局与物体关系。场景图提供了天然的结构先验，但以往基于图的方法多依赖复杂的GAN或扩散模型，速度和质量均落后于现代自回归架构。本文提出SATURN（用于统一渲染网络的三元组结构排列），作为VAR-CLIP的轻量级扩展，将场景图转化为重要性排序的令牌序列，使冻结的CLIP-VQ-VAE主干可解析图结构，仅微调VAR变换器。在Visual Genome数据集上，SATURN将FID从56.45降至21.62，Inception Score从16.03提升至24.78，优于SG2IM和SGDiff等方法，且无需额外模块或多阶段训练。定性结果显示物体数量和空间关系准确率显著提升，证明SATURN有效融合结构感知与前沿自回归保真度。
原文摘要 · Abstract (English)
State-of-the-art text-to-image models excel at photorealistic rendering but often struggle to capture the layout and object relationships implied by complex prompts. Scene graphs provide a natural structural prior, yet previous graph-guided approaches have typically relied on heavy GAN or diffusion pipelines, which lag behind modern autoregressive architectures in both speed and fidelity. We introduce SATURN (Structured Arrangement of Triplets for Unified Rendering Networks), a lightweight extension to VAR-CLIP that translates a scene graph into a salience-ordered token sequence, enabling a frozen CLIP-VQ-VAE backbone to interpret graph structure while fine-tuning only the VAR transformer. On the Visual Genome dataset, SATURN reduces FID from 56.45% to 21.62% and increases the Inception Score from 16.03 to 24.78, outperforming prior methods such as SG2IM and SGDiff without requiring extra modules or multi-stage training. Qualitative results further confirm improvements in object count fidelity and spatial relation accuracy, showing that SATURN effectively combines structural awareness with state-of-the-art autoregressive fidelity.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。