search

mattpoggi / mono-uncertainty

CVPR 2020 - On the uncertainty of self-supervised monocular depth estimation
237 stars 24 forks source link

some questions about my implementation #23

Closed Bowwowlol closed 2 years ago

Bowwowlol commented 2 years ago

Hi, @mattpoggi, Thanks for your excellent work!! I have some questions about my implementation.

  1. In the depth decoder (S), it is output 4 scale disparity and uncertainty, I would like to know that have you calculated all scale outputs when you implemented eq.14? image

  2. If my uncertainty has some negative values, Should I limit it to>0?

mattpoggi commented 2 years ago

Hi @Bowwowlol,

  1. depth and uncertainty at each scale are upsampled to full resolution when computing the loss
  2. if you assume your output uncertainty is u=log(uncertainty), you should be fine (then replace sigma with exp(u) in the loss function)
Bowwowlol commented 2 years ago

Hi, @mattpoggi, Thank you for the quick reply!! I really appreciate your help in resolving problems, That is very helpful to me. Good luck!!