LOBE-GS: Load-Balanced and Efficient 3D Gaussian Splatting for Large-Scale Scene Reconstruction

2025-10-02

AI Summary

To address scalability limitations, imbalanced memory consumption, and low training efficiency in 3D Gaussian Splatting (3DGS) reconstruction for large-scale, unbounded urban scenes, this paper proposes the first end-to-end optimized framework. Our method introduces three core innovations: (1) depth-aware spatial partitioning for geometry-adaptive scene decomposition; (2) visibility-optimized Gaussian load balancing to mitigate computational imbalance inherent in divide-and-conquer strategies; and (3) visibility-guided pruning and selective densification, which effectively reuse coarse-stage information while suppressing redundant Gaussian growth. Evaluated on large urban and complex outdoor scenes, our approach achieves a 2× speedup in end-to-end training over native 3DGS and state-of-the-art tiling-based methods, while preserving high-fidelity reconstruction quality. Notably, it is the first to enable native 3DGS–compatible reconstruction of square-kilometer–scale scenes—previously infeasible with conventional approaches.

Technology Category

• Computer Vision: 3D Computer Vision
• Search and Optimization: Distributed Search
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Security and Privacy: Large-scale security measurements

Abstract

3D Gaussian Splatting (3DGS) has established itself as an efficient representation for real-time, high-fidelity 3D scene reconstruction. However, scaling 3DGS to large and unbounded scenes such as city blocks remains difficult. Existing divide-and-conquer methods alleviate memory pressure by partitioning the scene into blocks, but introduce new bottlenecks: (i) partitions suffer from severe load imbalance since uniform or heuristic splits do not reflect actual computational demands, and (ii) coarse-to-fine pipelines fail to exploit the coarse stage efficiently, often reloading the entire model and incurring high overhead. In this work, we introduce LoBE-GS, a novel Load-Balanced and Efficient 3D Gaussian Splatting framework, that re-engineers the large-scale 3DGS pipeline. LoBE-GS introduces a depth-aware partitioning method that reduces preprocessing from hours to minutes, an optimization-based strategy that balances visible Gaussians -- a strong proxy for computational load -- across blocks, and two lightweight techniques, visibility cropping and selective densification, to further reduce training cost. Evaluations on large-scale urban and outdoor datasets show that LoBE-GS consistently achieves up to $2 imes$ faster end-to-end training time than state-of-the-art baselines, while maintaining reconstruction quality and enabling scalability to scenes infeasible with vanilla 3DGS.

Problem

• Achieving load balance in large-scale 3D Gaussian Splatting reconstruction
• Reducing preprocessing time and computational overhead in scene partitioning
• Enabling scalable high-fidelity 3D reconstruction for unbounded urban scenes

Innovation

• Depth-aware partitioning reduces preprocessing time
• Optimization balances computational load across blocks
• Visibility cropping and selective densification cut training cost