Depth-Regularized Optimization for 3D Gaussian Splatting in Few-Shot Images

TL;DR

This paper introduces a depth-regularized optimization technique for 3D Gaussian splatting that improves scene reconstruction quality from few images by incorporating a pre-trained depth map to prevent overfitting and enhance geometric accuracy.

Contribution

We propose a novel depth-guided optimization method for 3D Gaussian splatting that leverages pre-trained monocular depth estimation to improve results with limited images.

Findings

Enhanced geometric robustness over original methods

Effective mitigation of overfitting in few-shot scenarios

Improved visual quality and geometric consistency

Abstract

In this paper, we present a method to optimize Gaussian splatting with a limited number of images while avoiding overfitting. Representing a 3D scene by combining numerous Gaussian splats has yielded outstanding visual quality. However, it tends to overfit the training views when only a small number of images are available. To address this issue, we introduce a dense depth map as a geometry guide to mitigate overfitting. We obtained the depth map using a pre-trained monocular depth estimation model and aligning the scale and offset using sparse COLMAP feature points. The adjusted depth aids in the color-based optimization of 3D Gaussian splatting, mitigating floating artifacts, and ensuring adherence to geometric constraints. We verify the proposed method on the NeRF-LLFF dataset with varying numbers of few images. Our approach demonstrates robust geometry compared to the original method that relies solely on images. Project page: robot0321.github.io/DepthRegGS