Skip to content

Conversation

@AyushP123
Copy link
Collaborator

@as821, @hpvok13 we fixed a few bugs and got monodepth2 to run with PoseNet and DepthNet. However, we see negative losses. Can you proofread our changes?

@as821
Copy link
Owner

as821 commented Dec 3, 2022

@AyushP123 @adnamad
Seems reasonable from a quick look. Add assertions throughout the loss function computation to determine what term is making the loss go negative.

valmikikothare and others added 6 commits December 3, 2022 14:04
1. Removed 0.01 multiplication factor from PoseNet a and b.
2. Reprojection loss was getting scaled twice by sigma, removed the scecond scaling.
3. Transformed sigma while scaling reprojection loss to 10*sigma + 0.1 as mentioned
here -> https://github.com/no-Seaweed/Learning-Deep-Learning-1/blob/master/paper_notes/sfm_learner.md
4. Removed print statements and breakpoints
as821 added a commit that referenced this pull request Dec 8, 2022
* get trained posenet running, still have to fully integrate into backend

* perform brightness transform

* use sigma from depthnet

* run evaluations
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants