Point-to-Voxel Knowledge Distillation for LiDAR Semantic Segmentation

Open paper

Benchmark	Model	Rank	Results
3d-semantic-segmentation-on-semantickitti	PVKD	–	test mIoU: 71.2%
lidar-semantic-segmentation-on-nuscenes	Point-to-Voxel KD	–	val mIoU: 0.760