Question for the code

[SCUTykLin]

Hello:
What a great paper! I have two questions.
1) Can u provide the pre-trained encoder-decoder? I see that you seem to have used dataset pre-train the encoder and decoder.
2) Is it possible to train using only RGB? Is depth necessary? Is it possible to use only RGB in some simple scenarios?