Fails to load pretrained model
brahami14 opened this issue · comments
Can you change another version of spconv 2.x
?
thank you so much for your answer @X-Lai
@X-Lai I have another question please why did you choose different voxel size for semantic kitti and nuscenes ?
Their sparsities are totally different. SemanticKITTI scenes scanned by 64-line LiDAR, while nuscenes scenes are by 32-line LiDAR.