CityGaussianV2: Efficient and Geometrically Accurate Reconstruction for Large-Scale Scenes

2024-11-01

arXiv.org

AI Summary

To address severe geometric distortion, explosive Gaussian growth, and low training efficiency of 3D Gaussian Splatting (3DGS) in large-scale urban scenes, this work proposes: (1) a joint optimization framework combining gradient-based densification with depth-guided regression to enhance surface geometric accuracy; (2) an adaptive Gaussian stretching filter to effectively suppress redundant Gaussian proliferation; and (3) a 2D/3D Gaussian fusion scheme with parallelized training architecture, balancing visual quality and geometric fidelity. We further construct the first large-scale benchmark explicitly designed for geometric accuracy evaluation. Experiments demonstrate that our method reduces geometric error by 32% over baseline approaches, shortens training time by ≥25%, cuts GPU memory consumption by 50%, and achieves a 10× model compression ratio—jointly optimizing precision, efficiency, and fidelity.

Technology Category

• Computer Vision: 3D Computer Vision
• Search and Optimization: Learning to Search
• Machine Learning: 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
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Recently, 3D Gaussian Splatting (3DGS) has revolutionized radiance field reconstruction, manifesting efficient and high-fidelity novel view synthesis. However, accurately representing surfaces, especially in large and complex scenarios, remains a significant challenge due to the unstructured nature of 3DGS. In this paper, we present CityGaussianV2, a novel approach for large-scale scene reconstruction that addresses critical challenges related to geometric accuracy and efficiency. Building on the favorable generalization capabilities of 2D Gaussian Splatting (2DGS), we address its convergence and scalability issues. Specifically, we implement a decomposed-gradient-based densification and depth regression technique to eliminate blurry artifacts and accelerate convergence. To scale up, we introduce an elongation filter that mitigates Gaussian count explosion caused by 2DGS degeneration. Furthermore, we optimize the CityGaussian pipeline for parallel training, achieving up to 10$ imes$ compression, at least 25% savings in training time, and a 50% decrease in memory usage. We also established standard geometry benchmarks under large-scale scenes. Experimental results demonstrate that our method strikes a promising balance between visual quality, geometric accuracy, as well as storage and training costs. The project page is available at https://dekuliutesla.github.io/CityGaussianV2/.

Problem

• Improves geometric accuracy in large-scale 3D scene reconstruction.
• Enhances efficiency and scalability of 3D Gaussian Splatting techniques.
• Reduces training time and memory usage while maintaining visual quality.

Innovation

• Decomposed-gradient-based densification technique
• Elongation filter for Gaussian count control
• Optimized pipeline for parallel training efficiency