Search papers, labs, and topics across Lattice.
This paper introduces GeoScene, a novel framework that integrates satellite imagery and OpenStreetMap data to enhance 3D semantic scene completion by addressing the limitations of onboard image data. By learning voxel-wise reliability weights, GeoScene effectively balances local visual evidence with broader geospatial context, enabling improved feature refinement in both observed and unobserved areas. Experiments on SemanticKITTI and SSCBench-KITTI-360 show significant advancements in geometric and semantic completion, particularly for large-scale, structured classes, highlighting the effectiveness of geospatial priors in overcoming occlusions and limited fields of view.
GeoScene achieves remarkable improvements in 3D scene completion by leveraging satellite imagery and OpenStreetMap data, revealing the power of geospatial context in addressing onboard data limitations.
Inferring complete 3D geometry and semantics from onboard images remains challenging because occlusions and restricted fields of view leave large scene regions underconstrained. Although satellite imagery provides wide-area context, appearance cues alone offer limited structural guidance and may be unreliable because of spatial or temporal discrepancies. We present GeoScene, a geospatially guided framework that jointly uses satellite imagery and structured OpenStreetMap cues as soft priors for 3D semantic scene completion. GeoScene learns complementary voxel-wise reliability weights for onboard observations and geospatial guidance, and uses them to control feature refinement in observed and unobserved regions. This design preserves local visual evidence while exploiting large-scale road and building structure beyond onboard visibility. Experiments on SemanticKITTI and SSCBench-KITTI-360 demonstrate that GeoScene consistently improves both geometric and semantic completion under the geospatial-prior-assisted setting, with the most pronounced benefits for large-scale static and geospatially structured classes.
