search

Cc-Hy / CMKD

Cross-Modality Knowledge Distillation Network for Monocular 3D Object Detection (ECCV 2022 Oral)
Apache License 2.0
107 stars 9 forks source link

Generation of depth GT #20

Closed charmeleonz closed 1 year ago

charmeleonz commented 1 year ago

Hi, thanks for the great work.

Just wondering how the dense depth GT used in the image branch is generated? According to CaDDN, it is obtained by projection from LiDAR followed by depth completion. Is depth completion actually performed?

Cc-Hy commented 1 year ago

Hello, The dense depth maps we use are provided by the official CaDDN/OpenPCDet. And according to CaDDN, the depth completion is performed. However, there are also some methods to directly use the sparse depth maps for supervision. For example, you can use min pooling operation or sparse average pooling to obtain the depth supervision labels.