LOBE-GS: Load-Balanced and Efficient 3D Gaussian Splatting for Large-Scale Scene Reconstruction

LOBE-GS re-engineers large-scale 3D Gaussian Splatting with depth-aware camera selection, load-balanced scene partitioning, visibility cropping, and selective densification. The pipeline reduces preprocessing from hours to minutes and reaches up to 2× faster end-to-end training while preserving reconstruction quality.