Search papers, labs, and topics across Lattice.
This paper introduces GaussianSeed, a hierarchical Gaussian occupancy prediction framework designed to enhance high-resolution 3D occupancy prediction for applications in autonomous driving and robotic navigation. By organizing scene primitives into a coarse-to-fine hierarchy, GaussianSeed effectively mitigates memory bottlenecks, achieving a spatial resolution of $0.1\text{m}$ while ensuring real-time inference. Extensive evaluations on the newly constructed TJScenes dataset and Occ3D-nuScenes show that GaussianSeed not only reduces latency but also competes strongly on accuracy, pushing the boundaries of efficiency in 3D scene representation.
Achieving $0.1\text{m}$ spatial resolution with real-time inference, GaussianSeed redefines the efficiency-quality balance in 3D occupancy prediction.
Vision-centric 3D occupancy prediction provides dense scene representations essential for autonomous driving and robotic navigation, yet existing methods struggle to scale to high voxel resolutions due to prohibitive computational costs. To address this, we introduce GaussianSeed, a progressive multi-scale Gaussian occupancy prediction framework that organizes primitives into a coarse-to-fine hierarchy. Benefiting from this hierarchical design, GaussianSeed effectively circumvents the memory bottlenecks inherent in dense representations, successfully scaling to a $0.1\text{m}$ spatial resolution while maintaining real-time inference capabilities. To comprehensively evaluate high-resolution geometric perception, we further construct TJScenes, a panoramic six-camera occupancy dataset with highly detailed $0.1\text{m}$ annotations. Extensive experiments on Occ3D-nuScenes and TJScenes demonstrate that GaussianSeed delivers the lowest latency among all evaluated methods while maintaining highly competitive accuracy, advancing the efficiency-quality frontier of high-resolution 3D occupancy prediction.
