Is the v0.5 pretrained checkpoint the one that supports both monocular depth prediction and depth completion?

[weikaih04]

Hi — I noticed the first version has two checkpoints: a pretrained one for monocular depth estimation, and a post-trained one for depth completion. But for the 0.5 ckpt, we only have the pretrained version. Does that mean this checkpoint supports both monocular depth estimation and depth completion?

[IceTTTb]

Hi @weikaih04
The v0.5 ckpt supports both depth refinement and depth completion from RGBD input.