You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It is the GT depth, and it is optional and only used for validation. You can also use photometric loss for validation by just removing "--with-gt" in the training scripts.
Besides, I suggest you using our new implementation (https://github.com/JiawangBian/sc_depth_pl). Where you can set "--val_mode photo", so you do not need .npy file for validation.
After making my own dataset, I don't know how to generate the npy for the corresponding image file, so I would like to ask for some advice?
The text was updated successfully, but these errors were encountered: