SceneMaker：去遮挡与姿态估计模型解耦的开放集三维场景生成
SceneMaker: Open-set 3D Scene Generation with Decoupled De-occlusion and Pose Estimation Model
把去遮挡与位姿估计从三维物体生成中解耦，并借助图像数据集扩充开放集遮挡模式，为开放集场景合成中「先补全再摆位」的流水线提供了可复用的模块化设计；其统一全局/局部自注意力与交叉注意力的位姿估计模型可直接迁移到物体插入、场景编辑等需要精确 6DoF 摆放的任务。
原文依据 we first decouple the de-occlusion model from 3D object generation
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论本文提出一种名为 SceneMaker 的解耦式三维场景生成框架。由于缺乏足够的开放集去遮挡与姿态估计先验，现有方法在严重遮挡与开放集设定下难以同时生成高质量几何与准确姿态。为解决这些问题，我们首先将去遮挡模型与三维物体生成解耦，并利用图像数据集与收集的去遮挡数据集对其进行增强，以获得更加多样的开放集遮挡模式。随后，我们提出一个统一的姿态估计模型，在自注意力与交叉注意力中同时整合全局与局部机制，以提升精度。此外，我们构建了一个开放集三维场景数据集，以进一步扩展姿态估计模型的泛化能力。大量实验表明，我们的解耦框架在室内场景与开放集场景上均具有优越性。代码与数据集已发布于 https://idea-research.github.io/SceneMaker/。
英文原文摘要 1003 字符
We propose a decoupled 3D scene generation framework called SceneMaker in this work. Due to the lack of sufficient open-set de-occlusion and pose estimation priors, existing methods struggle to simultaneously produce high-quality geometry and accurate poses under severe occlusion and open-set settings. To address these issues, we first decouple the de-occlusion model from 3D object generation, and enhance it by leveraging image datasets and collected de-occlusion datasets for much more diverse open-set occlusion patterns. Then, we propose a unified pose estimation model that integrates global and local mechanisms for both self-attention and cross-attention to improve accuracy. Besides, we construct an open-set 3D scene dataset to further extend the generalization of the pose estimation model. Comprehensive experiments demonstrate the superiority of our decoupled framework on both indoor and open-set scenes. Our codes and datasets is released at https://idea-research.github.io/SceneMaker/.