MCG-NJU/SparseOcc

Sensors used during inference

Hong753 opened this issue · 3 comments

Hi, it seems that you are using the function sampling_4d which requires lidar2img transformations during evaluation as well.
It is my understanding that you proposed a method that uses only camera images during inference.
Please let me know if there's been a misunderstanding.

Yes, the projection matrix is also required.

Issue is resolved. I confused the lidar point cloud with the actual lidar device.

We do not require lidar, but the projection matrix is still necessary