Gaussian Mapping for Evolving Scenes

2025-06-07

AI Summary

To address the challenges of sustained update, poor geometric-semantic consistency, and limited adaptability to long-term dynamic scenes (e.g., out-of-view object appearance/disappearance and structural evolution) in 3D Gaussian Splatting mapping, this paper proposes the first online-evolving, high-fidelity 3D Gaussian mapping system. Methodologically, we introduce a dynamic-scene-adaptive optimization framework integrating timeliness-aware keyframe management, cross-temporal observation consistency constraints, and incremental Gaussian parameter refinement. Unlike static or short-horizon dynamic approaches, our system robustly models prolonged scene evolution while preserving historical information and enabling real-time, jointly geometrically and semantically consistent reconstruction and novel-view synthesis. Extensive evaluations on both synthetic and real-world datasets demonstrate significant improvements in reconstruction accuracy and rendering quality over existing state-of-the-art methods.

Technology Category

• Computer Vision: 3D Computer Vision
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Mapping systems with novel view synthesis (NVS) capabilities are widely used in computer vision, with augmented reality, robotics, and autonomous driving applications. Most notably, 3D Gaussian Splatting-based systems show high NVS performance; however, many current approaches are limited to static scenes. While recent works have started addressing short-term dynamics (motion within the view of the camera), long-term dynamics (the scene evolving through changes out of view) remain less explored. To overcome this limitation, we introduce a dynamic scene adaptation mechanism that continuously updates the 3D representation to reflect the latest changes. In addition, since maintaining geometric and semantic consistency remains challenging due to stale observations disrupting the reconstruction process, we propose a novel keyframe management mechanism that discards outdated observations while preserving as much information as possible. We evaluate Gaussian Mapping for Evolving Scenes (GaME) on both synthetic and real-world datasets and find it to be more accurate than the state of the art.

Problem

• Addressing long-term dynamic scene changes in 3D mapping
• Maintaining geometric and semantic consistency in evolving scenes
• Improving accuracy in dynamic scene reconstruction and synthesis

Innovation

• Dynamic scene adaptation updates 3D representation
• Keyframe management discards outdated observations
• 3D Gaussian Splatting for evolving scenes