PoseGaussian：基于稀疏视角物体级三维高斯泼溅的未见物体 6D 姿态估计
PoseGaussian: 6D Pose Estimation for Unseen Objects via Sparse-View Object-Level 3D Gaussian Splatting
利用稀疏 RGB-D 视图构建物体级 3DGS 表示，并通过深度先验初始化、自适应密度控制与视图扭曲增强来抑制浮空伪影和外观过拟合，为无 CAD 模型的 6D 位姿估计提供可复用的重建-配准流程。
原文依据 we use sparse RGB-D views to inject a depth structure prior into the 3DGS initialization for stable structure
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论6D 姿态估计是计算机视觉和机器人操作中的一项关键技术。然而，许多方法仍然严重依赖难以获取的 CAD 模型。物体级三维重建提供了一条替代路径，而三维高斯泼溅（3DGS）凭借其训练和渲染效率展现出令人信服的潜力。然而，在稀疏参考视角下，3DGS 容易产生浮空伪影和外观过拟合，从而削弱姿态估计的稳定性。我们提出 PoseGaussian，一种基于改进 3DGS 的稀疏视角未见物体 6D 姿态估计方法。首先，我们使用稀疏 RGB-D 视角将深度结构先验注入 3DGS 初始化以获得稳定的结构，并采用自适应密度控制、视图扭曲增强以及联合光度-深度监督来减少稀疏参考视角下的浮空伪影和外观过拟合。接下来，在姿态估计阶段，我们应用一种两阶段学习引导的 ICP 初始化器，利用几何特征获得稳定的初始姿态。最后，我们引入一种基于 3DGS 的迭代姿态细化器，在 appearance 和几何上对齐渲染图像与查询图像，进一步提高姿态估计精度。在 LINEMOD、GenMOP 以及我们的真实世界数据集上的实验表明，PoseGaussian 在无模型和稀疏视角设置下相比基线方法取得了显著改进，展现出对未见物体的强泛化能力和对视角稀疏性的鲁棒性。
英文原文摘要 1515 字符
6D pose estimation is a key technology in computer vision and robotic manipulation. However, many methods remain heavily dependent on CAD models that are difficult to obtain. Object-level 3D reconstruction provides an alternative route, and 3D Gaussian Splatting (3DGS) shows convincing potential owing to its training and rendering efficiency. Nevertheless, under sparse reference views, 3DGS is prone to floating artifacts and appearance overfitting, which weakens the stability of pose estimation. We present PoseGaussian, a method for sparse-view 6D pose estimation for unseen objects that builds on improved 3DGS. First, we use sparse RGB-D views to inject a depth structure prior into the 3DGS initialization for stable structure, and we adopt adaptive density control, view-warping augmentation, and joint photometric-depth supervision to reduce floaters and appearance overfitting under sparse reference views. Next, in the pose estimation stage, we apply a two-stage learning-guided ICP initializer that exploits geometric features to obtain a stable initial pose. Finally, we introduce a 3DGS-based iterative pose refiner that aligns rendered and query images in both appearance and geometry, further improving pose estimation accuracy. Experiments on LINEMOD, GenMOP, and our real-world datasets show that PoseGaussian achieves significant improvements over baseline methods under model-free and sparse-view settings, demonstrating strong generalization to unseen objects and robustness to view sparsity.