You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Oct 7, 2024. It is now read-only.
Hi, Thank you for the great work.
I have tried SGM implementations like OpenCV StereoSGBM_create or libSGM. But unfortunately none of them can reach the D1 & EPE indicated in your paper (D1=4.01, EPE=1.00 in Table 1, SGM with LRC).
Also, nearly all SGM implementations require hyper-parameters like P1, P2, uniqueness. It will be highly appreciated if you could provide yours when tuning on KITTI 2015 trainset.
Best.
The text was updated successfully, but these errors were encountered:
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Hi, Thank you for the great work.
I have tried SGM implementations like OpenCV
StereoSGBM_create
or libSGM. But unfortunately none of them can reach the D1 & EPE indicated in your paper (D1=4.01, EPE=1.00 in Table 1, SGM with LRC).Also, nearly all SGM implementations require hyper-parameters like P1, P2, uniqueness. It will be highly appreciated if you could provide yours when tuning on KITTI 2015 trainset.
Best.
The text was updated successfully, but these errors were encountered: