Momentum-GS: Momentum Gaussian Self-Distillation for High-Quality Large Scene Reconstruction

TL;DR

Momentum-GS introduces a momentum-based self-distillation approach for large-scale scene reconstruction, effectively improving accuracy and consistency while reducing memory and storage requirements, and decoupling block number from GPU count.

Contribution

The paper proposes a novel momentum-based self-distillation method that enhances block-wise training for large scene reconstruction, addressing data diversity and GPU limitations.

Findings

Achieves 12.8% improvement in LPIPS over CityGaussian.

Outperforms existing techniques on large-scale scenes.

Allows flexible block division independent of GPU count.

Abstract

3D Gaussian Splatting has demonstrated notable success in large-scale scene reconstruction, but challenges persist due to high training memory consumption and storage overhead. Hybrid representations that integrate implicit and explicit features offer a way to mitigate these limitations. However, when applied in parallelized block-wise training, two critical issues arise since reconstruction accuracy deteriorates due to reduced data diversity when training each block independently, and parallel training restricts the number of divided blocks to the available number of GPUs. To address these issues, we propose Momentum-GS, a novel approach that leverages momentum-based self-distillation to promote consistency and accuracy across the blocks while decoupling the number of blocks from the physical GPU count. Our method maintains a teacher Gaussian decoder updated with momentum, ensuring a stable reference during training. This teacher provides each block with global guidance in a self-distillation manner, promoting spatial consistency in reconstruction. To further ensure consistency across the blocks, we incorporate block weighting, dynamically adjusting each block's weight according to its reconstruction accuracy. Extensive experiments on large-scale scenes show that our method consistently outperforms existing techniques, achieving a 12.8% improvement in LPIPS over CityGaussian with much fewer divided blocks and establishing a new state of the art. Project page: https://jixuan-fan.github.io/Momentum-GS_Page/