Search papers, labs, and topics across Lattice.
FPSGen introduces a novel framework for point cloud scene generation that operates independently of partial scans, addressing the limitations of existing methods that rely on LiDAR data. By predicting a bird's-eye-view (BEV) prior and employing a teacher-student optimal transport scheme, FPSGen enables the generation of complete and coherent scenes even in occluded areas. The framework achieves state-of-the-art performance on SemanticKITTI and excels in unconditional generation on KITTI-360, demonstrating its flexibility and robustness in various scenarios.
Achieving state-of-the-art scene completion and generation without relying on partial scans, FPSGen redefines how point clouds can be generated in challenging environments.
Existing point-based generative methods for outdoor scenes primarily focus on LiDAR-conditioned completion. During training, noisy point clouds are constructed by perturbing complete ground-truth scenes, whereas during inference, they are initialized by adding noise to duplicated partial scans. This train-inference mismatch inherits the sparsity and visibility bias of partial scans, leading to sparse distant regions and incomplete geometry in occluded areas. Moreover, the reliance on partial scans restricts generation when LiDAR observations are unavailable or replaced by layout cues. We present FPSGen, a flexible framework that constructs point sources independently of partial scans. FPSGen first predicts a bird's-eye-view (BEV) prior with density, height, and mask channels from the active cues. The density map is then sampled to form a BEV-supported point source, enabling both unconditional and conditioned initialization. A teacher-student approximate optimal transport scheme then uses teacher-predicted endpoints to learn a velocity field that induces straighter transport paths. By integrating BEV point source construction with path-straightening transport, FPSGen provides a unified framework for unconditional and flexible cue-conditioned scene generation. Extensive experiments show that FPSGen achieves state-of-the-art JSD and voxel IoU performance on SemanticKITTI completion while maintaining strong performance with a single point transport step. On KITTI-360 unconditional generation, it also achieves the best Coverage (COV) among the compared methods.
