Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Evaluation Clipping #28

Open
Davidyao99 opened this issue Oct 14, 2024 · 1 comment
Open

Evaluation Clipping #28

Davidyao99 opened this issue Oct 14, 2024 · 1 comment

Comments

@Davidyao99
Copy link

I was looking through the evaluation script and was wondering why do we perform this clipping of depth values here

I was wondering if we can perform clipping of predicted depth values before the alignment stage. I notice that for my prediction, there were some extreme outlier values that led to poor shift / scale alignment values that led to extremely poor results for that particular scene. I was wondering if you have any advice / guidance for such an issue?

Thank you! and great work!

@gaoxiangjun
Copy link
Collaborator

Since we need to calculate relative depth through 1/disp, in some cases, we found extremely low disp would result in a huge depth, which would influence the align stage and tend to align more towards the large depth. Furthermore, this led to poor shift/scale alignment values.

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

No branches or pull requests

2 participants