How do you create the `depth`, `mask` and `mask_visib` from the `rgb` image?

[meropis]

I have an rgb image I wish to try and run the program on, with an image I found off the internet (it should work since the image I found contains one of the items the algorithm was trained on). How would I go about creating the mask_visib, mask and depth images from my rgb image?

[rasmushaugaard]

You actually don't need any of those. You do need a crop though. Perhaps the inference script can help you out.