Closed JuewenPeng closed 4 years ago
Hi, the model used for depth refinement in our paper is trained by attaching on the head of a monocular depth estimation model(SharpNet specifically) for all different depth estimation methods, which produces results with higher numerical results. Models trained with this repo are a bit lower numerically, but visually much better, and simpler. Hope it helps.
Thanks for your reply.
Hi, in your paper, is the refinement model trained on different depth estimation methods separately, or done in the same way with this repo?