Open
Description
Hello great work @AlessioTonioni and team!
I am trying to use the work for training on custom data which has spare disparity maps (16 bit) as shown below:
I did not modify any code. However, as the training progresses, the disparity predicted by the network has always been like this: (Except for the first frame)
Do you know why? Can you help me?
Metadata
Metadata
Assignees
Labels
No labels