PatchScene: Patch-based Voxel Diffusion for Large-Scale Scene Completion
This work addresses the challenge of scalable and coherent 3D scene completion for autonomous driving, offering a method that outperforms prior art in both accuracy and temporal consistency.
PatchScene introduces a patch-based voxel diffusion framework for large-scale LiDAR scene completion, achieving state-of-the-art performance on SemanticKITTI across all metrics and generalizing from 20m to 50m ranges without retraining.
We propose PatchScene, a novel diffusion-based framework for large-scale LiDAR scene completion. Unlike existing methods that rely on global latent representations or dense voxel grids, PatchScene adopts a patch-based voxel diffusion paradigm that explicitly generates fine-grained geometry within localized 3D regions. To ensure coherent reconstruction at both spatial and temporal scales, we introduce a confidence-guided spatio-temporal fusion mechanism that integrates overlapping patches and adjacent frames in a unified generative process. Furthermore, we design an Annular-Flow diffusion strategy that leverages the radial density pattern of LiDAR scans to progressively propagate high-fidelity information from near-range to far-range regions, enabling spatially unbounded scene completion. Extensive experiments on the SemanticKITTI benchmark demonstrate that PatchScene achieves state-of-the-art performance across all standard metrics, surpassing previous approaches in both geometric accuracy and temporal consistency. Remarkably, the model trained on 20 m LiDAR ranges generalizes effectively to 50 m scenes without retraining, highlighting its strong scalability and generalization capability for real-world autonomous driving applications.