解决密集重叠场景下生成物体数量不准的问题
Learning to Generate Multiple Objects from Dense and Occluded Layouts
- 引入布局感知注意力，抑制重叠物体间的干扰
- 使用非可见性平衡损失，提升被遮挡物体的训练信号
- 构建4.5万样本基准数据集，支持系统评估
文本到图像的扩散模型在密集场景中生成物体数量不准确，重叠实例会坍缩为难以区分的结构，尽管外观看似合理。我们识别出这一现象为实例归属坍缩：重叠物体的令牌在注意力机制中自由交互，而被严重遮挡的实例因可见区域小，接收的监督信号弱。为此，我们提出布局感知注意力偏置，软性引导令牌交互向区域一致分组，并抑制跨实例泄露；同时设计非可见性平衡损失，根据遮挡程度放大被遮挡物体的梯度。为支持系统评估，我们构建了包含4.5万张密集重叠场景的OverlapDepth-45K基准，带非可见性标注。所提方法显著提升计数准确性，防止实例合并，同时保持图像质量。
原文摘要 · Abstract (English)
Text-to-image diffusion models fail to generate correct object counts in dense scenes, where overlapping instances collapse into indistinguishable structures despite appearing visually plausible. We identify this as instance ownership collapse: tokens from overlapping objects interact freely through attention, while heavily occluded instances receive weak supervision due to their small visible areas. We address this through layout-aware attention biases that softly bias token interactions toward region-consistent grouping and suppress cross-instance leakage, paired with an amodal-balanced loss that amplifies gradients for occluded objects based on their occlusion level. To enable systematic evaluation, we introduce OverlapDepth-45K, a benchmark of densely overlapping scenes with amodal supervision. Our approach substantially improves count accuracy and prevents instance merging while preserving image quality. Project page: https://bachngoh.github.io/AIBL
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。