SplArt：基于3D高斯溅射的关节估计与部件级重建
SplArt: Articulation Estimation and Part-Level Reconstruction with 3D Gaussian Splatting
AI总结：
针对现有关节物体重建方法的可扩展性、鲁棒性及渲染缺陷，提出自监督类别无关框架SplArt，基于3D高斯溅射结合可微分移动性参数与多阶段优化，仅需两组不同关节状态的带位姿RGB图像即可实现部件级重建与关节估计，性能达当前最优且具备实用价值。
摘要：
重建日常环境中普遍存在的关节物体，对增强/虚拟现实及机器人技术应用至关重要。然而现有方法存在可扩展性局限（需要3D监督或成本高昂的标注）、鲁棒性问题（易陷入局部最优）以及渲染缺陷（速度不足或缺乏真实感）。我们提出SplArt，这是一种自监督、类别无关的框架，利用3D Gaussian Splatting（3DGS，3D高斯溅射），从两组处于不同关节状态的带位姿RGB图像中重建关节物体并推断运动学特性，可针对新视角和新关节状态实现实时真实感渲染。SplArt为每个3D高斯添加可微分移动性参数，实现精细的部件分割。框架采用多阶段优化策略，逐步处理重建、部件分割与关节估计任务，显著提升了鲁棒性与准确性。SplArt利用几何自监督，无需3D标注或类别特定先验，即可有效应对挑战性场景。在既有基准与新提出的基准上的评估，以及使用手持RGB相机开展的真实场景应用测试，均证明SplArt达到了当前最优性能且具备实际应用价值。代码已公开于https://github.com/ripl/splart。
英文摘要：
Reconstructing articulated objects prevalent in daily environments is crucial for applications in augmented/virtual reality and robotics. However, existing methods face scalability limitations (requiring 3D supervision or costly annotations), robustness issues (being susceptible to local optima), and rendering shortcomings (lacking speed or photorealism). We introduce SplArt, a self-supervised, category-agnostic framework that leverages 3D Gaussian Splatting (3DGS) to reconstruct articulated objects and infer kinematics from two sets of posed RGB images captured at different articulation states, enabling real-time photorealistic rendering for novel viewpoints and articulations. SplArt augments 3DGS with a differentiable mobility parameter per Gaussian, achieving refined part segmentation. A multi-stage optimization strategy is employed to progressively handle reconstruction, part segmentation, and articulation estimation, significantly enhancing robustness and accuracy. SplArt exploits geometric self-supervision, effectively addressing challenging scenarios without requiring 3D annotations or category-specific priors. Evaluations on established and newly proposed benchmarks, along with applications to real-world scenarios using a handheld RGB camera, demonstrate SplArt's state-of-the-art performance and real-world practicality. Code is publicly available at https://github.com/ripl/splart.
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。
