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
Hello:
Thanks for your open-source implementation! I run your code without modification and find the results on 14 classes are same,except on the class "transistor" (R18-Rd100):
image ROCAUC: 0.887
pixel ROCAUC: 0.959
vs : (yours)
image ROCAUC: 0.925
pixel ROCAUC: 0.968
I'm confused because i don't find randomness and all other classes's results are same. Do you have any idea about this?
btw, the per-pixel level ROCAUC seems to be calculated on all images once. How about evalute on single picture and average the AUC? Which is more reasonable?
The text was updated successfully, but these errors were encountered:
Hello:
Thanks for your open-source implementation! I run your code without modification and find the results on 14 classes are same,except on the class "transistor" (R18-Rd100):
image ROCAUC: 0.887
pixel ROCAUC: 0.959
vs : (yours)
image ROCAUC: 0.925
pixel ROCAUC: 0.968
I'm confused because i don't find randomness and all other classes's results are same. Do you have any idea about this?
btw, the per-pixel level ROCAUC seems to be calculated on all images once. How about evalute on single picture and average the AUC? Which is more reasonable?
The text was updated successfully, but these errors were encountered: