Search papers, labs, and topics across Lattice.
This paper introduces FocusGS, a novel framework for 3D scene reconstruction that enhances efficiency by focusing computational resources on areas of geometric ambiguity rather than uniformly processing the entire volume. By utilizing a 3D Geometric Ambiguity Manifold, FocusGS effectively identifies and targets regions with high uncertainty, leading to significant reductions in the number of Gaussians used and rendering times. Experimental results show that FocusGS not only improves the efficiency-quality trade-off but also sets new benchmarks in driving-centric 3D reconstruction tasks.
Targeted structural completion can slash Gaussian usage by 74% and rendering time by 34%, revolutionizing 3D reconstruction in autonomous driving.
Reconstructing 3D scene structures from sparse, low-overlap observations remains a fundamental challenge in autonomous driving. Recent state-of-the-art frameworks achieve promising results by incorporating voxel-based Gaussians, but incur substantial computational redundancy due to a uniform volumetric processing strategy. To bridge the gap between the efficiency of pixel-based Gaussian methods and the structural completeness of voxel-based Gaussian approaches, we propose FocusGS, a simple yet effective framework that shifts the paradigm from global densification to targeted structural completion. Our central insight is that structural completion should be decoupled from deterministic regions, with computation concentrated exclusively on areas exhibiting geometric ambiguity. Specifically, FocusGS addresses the localization challenge by deriving a 3D Geometric Ambiguity Manifold to accurately isolate localized areas prone to occlusion and high geometric uncertainty. To overcome the subsequent manifold completion challenge, we design a lightweight targeted structure completion module that selectively instantiates and optimizes continuous Gaussian queries strictly within this unstructured, sparse topological subspace. Extensive experiments demonstrate that FocusGS achieves a superior efficiency-quality trade-off, advancing state-of-the-art performance on driving-centric benchmarks while naturally reducing the total number of Gaussians by ~74% and decreasing rendering time by ~34%.
