OcclusionFormer: 布局导向图像生成中的Z轴顺序安排
OcclusionFormer: Arranging Z-Order for Layout-Grounded Image Generation
AI总结：
本文提出OcclusionFormer，一种基于Z轴顺序的扩散变换框架，通过解耦实例并利用体积渲染进行合成，以解决布局到图像模型中物体间遮挡问题，并通过查询对齐损失提升空间精度和语义一致性。
摘要：
最近的布局到图像模型在空间可控性方面取得了显著进展。然而，它们仍然在物体间遮挡方面存在困难。当边界框重叠时，大多数现有方法缺乏显式的遮挡信息，这使得交集区域的生成本质上具有歧义性，并阻碍了复杂遮挡关系的确定。为此，我们首先构建了SA-Z，一个包含显式遮挡顺序和像素级注释的大型数据集。基于我们提出的数据集，我们引入了OcclusionFormer，一种新的遮挡感知扩散变换框架，通过解耦实例并利用体积渲染进行合成，显式地建模Z轴优先级。此外，为了确保细粒度的空间精度，我们引入了查询对齐损失，显式监督单个实例并增强语义一致性。所提出的方法有效减少了重叠区域的歧义性，强制正确遮挡依赖关系，并保持了结构完整性，从而在多样化的场景中实现了显著的准确性提升。
英文摘要：
Recent layout-to-image models have achieved remarkable progress in spatial controllability. However, they still struggle with inter-object occlusion. When bounding boxes overlap, most existing methods lack explicit occlusion information, which makes the generation in intersection regions inherently ambiguous and hinders the determination of complex occlusion relationships. As a result, they often produce entangled textures or physically inconsistent layering in the overlapped areas. To address this issue, we first construct SA-Z, a large-scale dataset enriched with explicit occlusion ordering and pixel-level annotations. Building upon our proposed dataset, we introduce OcclusionFormer, a novel occlusion-aware Diffusion Transformer framework that explicitly models Z-order priority by decoupling instances and compositing them via volume rendering. Furthermore, to ensure fine-grained spatial precision, we introduce a queried alignment loss that explicitly supervises individual instances and enhances semantic consistency. The proposed method effectively reduces ambiguity in overlapping regions, enforces correct occlusion dependencies, and preserves structural integrity, leading to substantial accuracy gains across diverse scenes.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
