Hey @Xharlie :
Thanks for the released code. I'm focusing on how the BtcDet is implemented.
I found a conflict in the generation of spherical voxels between code and paper.
In the code, I found that the point_cloud_range is [2.24, -40.69, -2.6, 69.12, 40.69, 0.64], voxel_size=[0.32, 0.52, 0.36]
but in your paper, the point_cloud_range is [2.24, -40.69, -16.60, 70.72, 40.69, 4.00], voxel_size=[0.32, 0.52, 0.42]
It seems that the range in x and z dimension are absolutely different(especially for z dimension).
Can you explain why such difference appears or it is just a clerical error?
Hey @Xharlie : Thanks for the released code. I'm focusing on how the BtcDet is implemented. I found a conflict in the generation of spherical voxels between code and paper. In the code, I found that the point_cloud_range is [2.24, -40.69, -2.6, 69.12, 40.69, 0.64], voxel_size=[0.32, 0.52, 0.36] but in your paper, the point_cloud_range is [2.24, -40.69, -16.60, 70.72, 40.69, 4.00], voxel_size=[0.32, 0.52, 0.42] It seems that the range in x and z dimension are absolutely different(especially for z dimension).
Can you explain why such difference appears or it is just a clerical error?