发表机构
Pusan National University; Electronics and Telecommunications Research Institute (ETRI)（釜山国立大学; 电子和电信研究所）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究动态3D高斯表示的多变形建模问题，提出变形专家混合（MoDE）和动态高斯点云专家混合（MoE-GS）两种方法，通过不同集成约束实现多变形建模，为该领域提供替代策略并阐明集成约束对变形专家设计及行为的影响。
AI 中文摘要
由于现实世界运动的异质性和空间变化特性，动态场景重建仍然具有挑战性。尽管最近的3D高斯点云方法为动态新视图合成引入了多种变形公式，但每种方法通常在其表示中依赖单一变形模型，这限制了在不同动态场景中的鲁棒性。本文研究了在两种不同的集成约束下，动态3D高斯表示的多变形建模这一基本问题。从专家混合（MoE）的角度，将多变形建模视为在统一3D表示中组合多个专业变形模型的问题。首先介绍了变形专家混合（MoDE），通过联合优化将多个变形专家直接集成到可变形高斯点云管道中。专家在共享的规范高斯表示上操作，无需引入额外训练阶段或修改原始优化计划。还提出了动态高斯点云专家混合（MoE-GS），在不同集成约束下，变形专家独立优化并通过单独路由阶段组合；专家交互发生在个体优化后的非规范高斯表示上。这两种方法为多变形建模提供了替代策略，阐明了集成约束如何塑造动态3D高斯表示中变形专家的设计和行为。
英文摘要
Dynamic scene reconstruction remains challenging due to the heterogeneous and spatially varying nature of real-world motion. Although recent 3D Gaussian Splatting methods have introduced diverse deformation formulations for dynamic novel view synthesis, each method typically relies on a single deformation model within its representation, which limits robustness across diverse dynamic scenarios. In this work, we study a fundamental problem-multi-deformation modeling for dynamic 3D Gaussian representations-under two distinct integration constraints that differ in when and how multiple deformation experts interact during training. From a Mixture-of-Experts (MoE) perspective, we view multi-deformation modeling as the problem of combining multiple specialized deformation models within a unified 3D representation. We first introduce Mixture of Deformation Experts (MoDE), which integrates multiple deformation experts directly into the deformable Gaussian Splatting pipeline through joint optimization. In MoDE, experts operate on a shared canonical Gaussian representation, enabling multi-deformation modeling without introducing additional training stages or modifying the original optimization schedule. In contrast, we further present Mixture of Experts for Dynamic Gaussian Splatting (MoE-GS) under a different integration constraint, where deformation experts are optimized independently and combined through a separate routing stage. As a result, expert interaction occurs over non-canonical Gaussian representations after individual optimization. Together, these two approaches provide alternative strategies for multi-deformation modeling, clarifying how integration constraints shape the design and behavior of deformation experts in dynamic 3D Gaussian representations. Our code is available at: https://github.com/cvsp-lab/MoE-GS-studio.
CommentsIEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI, 2026)
